Warning: Permanently added '3.219.45.191' (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/8745035-fedora-rawhide-x86_64 --chroot fedora-rawhide-x86_64


Version: 1.1
PID: 9100
Logging PID: 9101
Task:
{'allow_user_ssh': False,
 'appstream': False,
 'background': True,
 'build_id': 8745035,
 'buildroot_pkgs': [],
 'chroot': 'fedora-rawhide-x86_64',
 'enable_net': False,
 'fedora_review': False,
 'git_hash': '73556b7995eff3b98b347a6ff71792fa1ac7b39e',
 'git_repo': 'https://copr-dist-git.fedorainfracloud.org/git/@rocm-packagers-sig/RH/llama-cpp',
 'isolation': 'default',
 'memory_reqs': 2048,
 'package_name': 'llama-cpp',
 'package_version': 'b4580-2',
 'project_dirname': 'RH',
 'project_name': 'RH',
 'project_owner': '@rocm-packagers-sig',
 'repo_priority': None,
 'repos': [{'baseurl': 'https://download.copr.fedorainfracloud.org/results/@rocm-packagers-sig/RH/fedora-rawhide-x86_64/',
            'id': 'copr_base',
            'name': 'Copr repository',
            'priority': None}],
 'sandbox': '@rocm-packagers-sig/RH--https://src.fedoraproject.org/user/trix',
 'source_json': {},
 'source_type': None,
 'ssh_public_keys': None,
 'storage': 0,
 'submitter': 'https://src.fedoraproject.org/user/trix',
 'tags': [],
 'task_id': '8745035-fedora-rawhide-x86_64',
 'timeout': 18000,
 'uses_devel_repo': False,
 'with_opts': [],
 'without_opts': []}

Running: git clone https://copr-dist-git.fedorainfracloud.org/git/@rocm-packagers-sig/RH/llama-cpp /var/lib/copr-rpmbuild/workspace/workdir-nlegjgrg/llama-cpp --depth 500 --no-single-branch --recursive

cmd: ['git', 'clone', 'https://copr-dist-git.fedorainfracloud.org/git/@rocm-packagers-sig/RH/llama-cpp', '/var/lib/copr-rpmbuild/workspace/workdir-nlegjgrg/llama-cpp', '--depth', '500', '--no-single-branch', '--recursive']
cwd: .
rc: 0
stdout: 
stderr: Cloning into '/var/lib/copr-rpmbuild/workspace/workdir-nlegjgrg/llama-cpp'...

Running: git checkout 73556b7995eff3b98b347a6ff71792fa1ac7b39e --

cmd: ['git', 'checkout', '73556b7995eff3b98b347a6ff71792fa1ac7b39e', '--']
cwd: /var/lib/copr-rpmbuild/workspace/workdir-nlegjgrg/llama-cpp
rc: 0
stdout: 
stderr: Note: switching to '73556b7995eff3b98b347a6ff71792fa1ac7b39e'.

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 73556b7 automatic import of llama-cpp

Running: dist-git-client sources

cmd: ['dist-git-client', 'sources']
cwd: /var/lib/copr-rpmbuild/workspace/workdir-nlegjgrg/llama-cpp
rc: 0
stdout: 
stderr: INFO: Reading stdout from command: git rev-parse --abbrev-ref HEAD
INFO: Reading stdout from command: git rev-parse HEAD
INFO: Reading sources specification file: sources
INFO: Downloading llama.cpp-b4580.tar.gz
INFO: Reading stdout from command: curl --help all
INFO: Calling: curl -H Pragma: -o llama.cpp-b4580.tar.gz --location --connect-timeout 60 --retry 3 --retry-delay 10 --remote-time --show-error --fail --retry-all-errors https://copr-dist-git.fedorainfracloud.org/repo/pkgs/@rocm-packagers-sig/RH/llama-cpp/llama.cpp-b4580.tar.gz/md5/5f83b7cd129f926b1a15dbbb65bc10af/llama.cpp-b4580.tar.gz
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 19.5M  100 19.5M    0     0  26.8M      0 --:--:-- --:--:-- --:--:-- 26.8M
INFO: Reading stdout from command: md5sum llama.cpp-b4580.tar.gz

/usr/bin/tail: /var/lib/copr-rpmbuild/main.log: file truncated
Running (timeout=18000): unbuffer mock --spec /var/lib/copr-rpmbuild/workspace/workdir-nlegjgrg/llama-cpp/llama-cpp.spec --sources /var/lib/copr-rpmbuild/workspace/workdir-nlegjgrg/llama-cpp --resultdir /var/lib/copr-rpmbuild/results --uniqueext 1741555222.965739 -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-nlegjgrg/llama-cpp/llama-cpp.spec --sources /var/lib/copr-rpmbuild/workspace/workdir-nlegjgrg/llama-cpp --resultdir /var/lib/copr-rpmbuild/results --uniqueext 1741555222.965739 -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-nlegjgrg/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-1741555222.965739/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-ebb5a5fe-976a-4699-8c88-946dca6257a1
INFO: Checking that 46175bb9093f653d5a8191d43121284a83f02d13d61cf4c00598759268be8f4c image matches host's architecture
INFO: Copy content of container 46175bb9093f653d5a8191d43121284a83f02d13d61cf4c00598759268be8f4c to /var/lib/mock/fedora-rawhide-x86_64-bootstrap-1741555222.965739/root
INFO: mounting 46175bb9093f653d5a8191d43121284a83f02d13d61cf4c00598759268be8f4c with podman image mount
INFO: image 46175bb9093f653d5a8191d43121284a83f02d13d61cf4c00598759268be8f4c as /var/lib/containers/storage/overlay/f0907060a18db94fc5a49359d6c448dee159119bf4d27d3c7e2d0fa04ac4c27a/merged
INFO: umounting image 46175bb9093f653d5a8191d43121284a83f02d13d61cf4c00598759268be8f4c (/var/lib/containers/storage/overlay/f0907060a18db94fc5a49359d6c448dee159119bf4d27d3c7e2d0fa04ac4c27a/merged) with podman image umount
INFO: Removing image mock-bootstrap-ebb5a5fe-976a-4699-8c88-946dca6257a1
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_64-1741555222.965739/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
  dnf5-5.2.11.0-1.fc43.x86_64
  dnf5-plugins-5.2.11.0-1.fc43.x86_64
Start: installing minimal buildroot with dnf5
Updating and loading repositories:
 fedora                                 100% |  40.7 MiB/s |  21.9 MiB |  00m01s
 Copr repository                        100% |   2.3 MiB/s | 153.1 KiB |  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 43-0.6                     fedora      20.1 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 2:4.17.0-4.fc42            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 1.31-3.fc42                fedora      66.2 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 8.12.1-1.fc43              fedora     457.2 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 6-2.fc42                   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 43-0.6                     fedora       0.0   B
 fedora-release-identity-basic     noarch 43-0.6                     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 3.7.7-3.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.fc42               fedora     841.3 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 8.12.1-1.fc43              fedora     850.1 KiB
 libeconf                          x86_64 0.7.6-1.fc43               fedora      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 15.0.1-0.9.fc43            copr_base  266.6 KiB
 libgomp                           x86_64 15.0.1-0.9.fc43            copr_base  535.9 KiB
 libidn2                           x86_64 2.3.7-3.fc42               fedora     329.0 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 15.0.1-0.9.fc43            copr_base    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 2.12.9-2.fc42              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 4.2.1-6.fc42               fedora     831.9 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 1:3.2.4-2.fc43             fedora       7.8 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 1.17.0-1.fc43              fedora       1.9 KiB
 python-srpm-macros                noarch 3.13-4.fc42                fedora      51.0 KiB
 qt5-srpm-macros                   noarch 5.15.15-1.fc42             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 2.15.0-13.fc43             fedora     720.9 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 52 MiB. Need to download 52 MiB.
After this operation, 176 MiB extra will be used (install 176 MiB, remove 0 B).
[  1/148] bzip2-0:1.0.8-20.fc42.x86_64  100% |   5.1 MiB/s |  52.1 KiB |  00m00s
[  2/148] coreutils-0:9.6-2.fc43.x86_64 100% |  60.0 MiB/s |   1.1 MiB |  00m00s
[  3/148] cpio-0:2.15-2.fc41.x86_64     100% |  31.7 MiB/s | 291.8 KiB |  00m00s
[  4/148] bash-0:5.2.37-3.fc43.x86_64   100% |  72.5 MiB/s |   1.8 MiB |  00m00s
[  5/148] fedora-release-common-0:43-0. 100% |   5.1 MiB/s |  25.9 KiB |  00m00s
[  6/148] diffutils-0:3.10-9.fc42.x86_6 100% |  56.4 MiB/s | 404.6 KiB |  00m00s
[  7/148] glibc-minimal-langpack-0:2.41 100% |  41.6 MiB/s | 127.9 KiB |  00m00s
[  8/148] findutils-1:4.10.0-5.fc42.x86 100% | 134.6 MiB/s | 551.5 KiB |  00m00s
[  9/148] grep-0:3.11-10.fc42.x86_64    100% |  97.7 MiB/s | 300.1 KiB |  00m00s
[ 10/148] gzip-0:1.13-3.fc42.x86_64     100% |  55.5 MiB/s | 170.4 KiB |  00m00s
[ 11/148] info-0:7.2-3.fc42.x86_64      100% |  59.8 MiB/s | 183.8 KiB |  00m00s
[ 12/148] patch-0:2.7.6-26.fc42.x86_64  100% |  41.8 MiB/s | 128.4 KiB |  00m00s
[ 13/148] redhat-rpm-config-0:342-2.fc4 100% |  39.9 MiB/s |  81.6 KiB |  00m00s
[ 14/148] rpm-build-0:4.20.1-1.fc43.x86 100% |  40.0 MiB/s |  81.8 KiB |  00m00s
[ 15/148] sed-0:4.9-4.fc42.x86_64       100% |  77.5 MiB/s | 317.3 KiB |  00m00s
[ 16/148] unzip-0:6.0-66.fc42.x86_64    100% |  45.1 MiB/s | 184.6 KiB |  00m00s
[ 17/148] tar-2:1.35-5.fc42.x86_64      100% | 105.3 MiB/s | 862.5 KiB |  00m00s
[ 18/148] shadow-utils-2:4.17.0-4.fc42. 100% | 131.0 MiB/s |   1.3 MiB |  00m00s
[ 19/148] which-0:2.23-1.fc42.x86_64    100% |  13.6 MiB/s |  41.7 KiB |  00m00s
[ 20/148] xz-1:5.6.3-3.fc42.x86_64      100% |  46.4 MiB/s | 474.9 KiB |  00m00s
[ 21/148] gawk-0:5.3.1-1.fc42.x86_64    100% | 107.9 MiB/s |   1.1 MiB |  00m00s
[ 22/148] util-linux-0:2.40.4-7.fc43.x8 100% |  96.1 MiB/s |   1.2 MiB |  00m00s
[ 23/148] filesystem-0:3.18-38.fc43.x86 100% | 166.4 MiB/s |   1.3 MiB |  00m00s
[ 24/148] glibc-0:2.41.9000-2.fc43.x86_ 100% | 207.3 MiB/s |   2.3 MiB |  00m00s
[ 25/148] ncurses-libs-0:6.5-5.20250125 100% |  32.7 MiB/s | 335.0 KiB |  00m00s
[ 26/148] bzip2-libs-0:1.0.8-20.fc42.x8 100% |   8.5 MiB/s |  43.6 KiB |  00m00s
[ 27/148] libacl-0:2.3.2-3.fc42.x86_64  100% |  11.2 MiB/s |  23.0 KiB |  00m00s
[ 28/148] gmp-1:6.3.0-3.fc43.x86_64     100% |  78.7 MiB/s | 322.2 KiB |  00m00s
[ 29/148] libattr-0:2.5.2-5.fc42.x86_64 100% |   5.6 MiB/s |  17.1 KiB |  00m00s
[ 30/148] coreutils-common-0:9.6-2.fc43 100% | 189.5 MiB/s |   2.1 MiB |  00m00s
[ 31/148] libcap-0:2.73-2.fc42.x86_64   100% |  13.7 MiB/s |  84.3 KiB |  00m00s
[ 32/148] libselinux-0:3.8-1.fc42.x86_6 100% |  15.8 MiB/s |  97.1 KiB |  00m00s
[ 33/148] fedora-repos-0:43-0.1.noarch  100% |   3.0 MiB/s |   9.3 KiB |  00m00s
[ 34/148] systemd-libs-0:257.4-3.fc43.x 100% | 130.9 MiB/s | 804.5 KiB |  00m00s
[ 35/148] glibc-common-0:2.41.9000-2.fc 100% |  81.1 MiB/s | 415.0 KiB |  00m00s
[ 36/148] openssl-libs-1:3.2.4-2.fc43.x 100% | 167.4 MiB/s |   2.3 MiB |  00m00s
[ 37/148] pcre2-0:10.45-1.fc43.x86_64   100% |  32.1 MiB/s | 262.8 KiB |  00m00s
[ 38/148] ed-0:1.21-2.fc42.x86_64       100% |  16.0 MiB/s |  82.0 KiB |  00m00s
[ 39/148] ansible-srpm-macros-0:1-17.1. 100% |   9.9 MiB/s |  20.3 KiB |  00m00s
[ 40/148] build-reproducibility-srpm-ma 100% |  11.4 MiB/s |  11.7 KiB |  00m00s
[ 41/148] efi-srpm-macros-0:6-2.fc42.no 100% |  11.0 MiB/s |  22.5 KiB |  00m00s
[ 42/148] dwz-0:0.15-9.fc42.x86_64      100% |  44.2 MiB/s | 135.7 KiB |  00m00s
[ 43/148] file-0:5.46-1.fc42.x86_64     100% |  15.9 MiB/s |  48.7 KiB |  00m00s
[ 44/148] filesystem-srpm-macros-0:3.18 100% |  12.5 MiB/s |  25.5 KiB |  00m00s
[ 45/148] fonts-srpm-macros-1:2.0.5-21. 100% |  13.2 MiB/s |  27.1 KiB |  00m00s
[ 46/148] forge-srpm-macros-0:0.4.0-2.f 100% |  19.4 MiB/s |  19.9 KiB |  00m00s
[ 47/148] fpc-srpm-macros-0:1.3-14.fc42 100% |   3.9 MiB/s |   8.0 KiB |  00m00s
[ 48/148] ghc-srpm-macros-0:1.9.2-2.fc4 100% |   4.5 MiB/s |   9.2 KiB |  00m00s
[ 49/148] gnat-srpm-macros-0:6-7.fc42.n 100% |   4.2 MiB/s |   8.6 KiB |  00m00s
[ 50/148] go-srpm-macros-0:3.6.0-6.fc42 100% |  13.5 MiB/s |  27.7 KiB |  00m00s
[ 51/148] kernel-srpm-macros-0:1.0-25.f 100% |   9.6 MiB/s |   9.9 KiB |  00m00s
[ 52/148] lua-srpm-macros-0:1-15.fc42.n 100% |   4.4 MiB/s |   8.9 KiB |  00m00s
[ 53/148] ocaml-srpm-macros-0:10-4.fc42 100% |   4.5 MiB/s |   9.2 KiB |  00m00s
[ 54/148] openblas-srpm-macros-0:2-19.f 100% |   3.8 MiB/s |   7.8 KiB |  00m00s
[ 55/148] package-notes-srpm-macros-0:0 100% |   4.5 MiB/s |   9.3 KiB |  00m00s
[ 56/148] perl-srpm-macros-0:1-57.fc42. 100% |   8.3 MiB/s |   8.5 KiB |  00m00s
[ 57/148] pyproject-srpm-macros-0:1.17. 100% |   6.8 MiB/s |  14.0 KiB |  00m00s
[ 58/148] python-srpm-macros-0:3.13-4.f 100% |  11.2 MiB/s |  23.0 KiB |  00m00s
[ 59/148] qt5-srpm-macros-0:5.15.15-1.f 100% |   4.3 MiB/s |   8.9 KiB |  00m00s
[ 60/148] qt6-srpm-macros-0:6.8.2-2.fc4 100% |   9.1 MiB/s |   9.3 KiB |  00m00s
[ 61/148] rust-srpm-macros-0:26.3-4.fc4 100% |   5.7 MiB/s |  11.7 KiB |  00m00s
[ 62/148] tree-sitter-srpm-macros-0:0.2 100% |   5.8 MiB/s |  11.9 KiB |  00m00s
[ 63/148] rpm-0:4.20.1-1.fc43.x86_64    100% | 132.7 MiB/s | 543.7 KiB |  00m00s
[ 64/148] zig-srpm-macros-0:1-4.fc42.no 100% |   4.0 MiB/s |   8.2 KiB |  00m00s
[ 65/148] zip-0:3.0-43.fc42.x86_64      100% |  85.8 MiB/s | 263.5 KiB |  00m00s
[ 66/148] debugedit-0:5.1-5.fc43.x86_64 100% |  38.4 MiB/s |  78.6 KiB |  00m00s
[ 67/148] elfutils-0:0.192-8.fc42.x86_6 100% | 107.6 MiB/s | 551.0 KiB |  00m00s
[ 68/148] elfutils-libelf-0:0.192-8.fc4 100% |  50.8 MiB/s | 208.1 KiB |  00m00s
[ 69/148] libarchive-0:3.7.7-3.fc43.x86 100% | 100.5 MiB/s | 411.6 KiB |  00m00s
[ 70/148] popt-0:1.19-8.fc42.x86_64     100% |  21.5 MiB/s |  66.0 KiB |  00m00s
[ 71/148] readline-0:8.2-13.fc43.x86_64 100% |  69.3 MiB/s | 212.9 KiB |  00m00s
[ 72/148] rpm-build-libs-0:4.20.1-1.fc4 100% |  32.5 MiB/s |  99.7 KiB |  00m00s
[ 73/148] rpm-libs-0:4.20.1-1.fc43.x86_ 100% |  76.2 MiB/s | 312.2 KiB |  00m00s
[ 74/148] zstd-0:1.5.7-1.fc43.x86_64    100% | 118.6 MiB/s | 485.8 KiB |  00m00s
[ 75/148] audit-libs-0:4.0.3-2.fc42.x86 100% |  30.6 MiB/s | 125.3 KiB |  00m00s
[ 76/148] libeconf-0:0.7.6-1.fc43.x86_6 100% |  17.2 MiB/s |  35.2 KiB |  00m00s
[ 77/148] libsemanage-0:3.8-1.fc42.x86_ 100% |  40.2 MiB/s | 123.6 KiB |  00m00s
[ 78/148] pam-libs-0:1.7.0-4.fc42.x86_6 100% |  28.5 MiB/s |  58.3 KiB |  00m00s
[ 79/148] libxcrypt-0:4.4.38-6.fc43.x86 100% |  31.1 MiB/s | 127.3 KiB |  00m00s
[ 80/148] setup-0:2.15.0-13.fc43.noarch 100% |  50.7 MiB/s | 155.8 KiB |  00m00s
[ 81/148] mpfr-0:4.2.1-6.fc42.x86_64    100% | 113.4 MiB/s | 348.5 KiB |  00m00s
[ 82/148] xz-libs-1:5.6.3-3.fc42.x86_64 100% |  27.7 MiB/s | 113.4 KiB |  00m00s
[ 83/148] libcap-ng-0:0.8.5-4.fc42.x86_ 100% |  15.7 MiB/s |  32.2 KiB |  00m00s
[ 84/148] libblkid-0:2.40.4-7.fc43.x86_ 100% |  39.9 MiB/s | 122.5 KiB |  00m00s
[ 85/148] libfdisk-0:2.40.4-7.fc43.x86_ 100% |  38.6 MiB/s | 158.2 KiB |  00m00s
[ 86/148] libmount-0:2.40.4-7.fc43.x86_ 100% |  50.5 MiB/s | 155.0 KiB |  00m00s
[ 87/148] libuuid-0:2.40.4-7.fc43.x86_6 100% |  12.4 MiB/s |  25.3 KiB |  00m00s
[ 88/148] libsmartcols-0:2.40.4-7.fc43. 100% |  19.8 MiB/s |  81.2 KiB |  00m00s
[ 89/148] zlib-ng-compat-0:2.2.4-2.fc43 100% |  25.8 MiB/s |  79.1 KiB |  00m00s
[ 90/148] util-linux-core-0:2.40.4-7.fc 100% | 129.3 MiB/s | 529.5 KiB |  00m00s
[ 91/148] ncurses-base-0:6.5-5.20250125 100% |  28.7 MiB/s |  88.1 KiB |  00m00s
[ 92/148] glibc-gconv-extra-0:2.41.9000 100% | 185.6 MiB/s |   1.7 MiB |  00m00s
[ 93/148] libsepol-0:3.8-1.fc42.x86_64  100% |  56.8 MiB/s | 348.9 KiB |  00m00s
[ 94/148] ca-certificates-0:2024.2.69_v 100% | 131.8 MiB/s | 945.0 KiB |  00m00s
[ 95/148] crypto-policies-0:20250305-1. 100% |  23.4 MiB/s |  95.8 KiB |  00m00s
[ 96/148] fedora-gpg-keys-0:43-0.1.noar 100% |  33.1 MiB/s | 135.6 KiB |  00m00s
[ 97/148] fedora-repos-rawhide-0:43-0.1 100% |   2.9 MiB/s |   8.8 KiB |  00m00s
[ 98/148] pcre2-syntax-0:10.45-1.fc43.n 100% |  52.6 MiB/s | 161.7 KiB |  00m00s
[ 99/148] add-determinism-0:0.6.0-1.fc4 100% | 149.5 MiB/s | 918.3 KiB |  00m00s
[100/148] curl-0:8.12.1-1.fc43.x86_64   100% |  54.8 MiB/s | 224.3 KiB |  00m00s
[101/148] file-libs-0:5.46-1.fc42.x86_6 100% | 138.2 MiB/s | 849.4 KiB |  00m00s
[102/148] elfutils-debuginfod-client-0: 100% |  22.7 MiB/s |  46.5 KiB |  00m00s
[103/148] elfutils-libs-0:0.192-8.fc42. 100% |  64.9 MiB/s | 265.9 KiB |  00m00s
[104/148] libzstd-0:1.5.7-1.fc43.x86_64 100% |  76.9 MiB/s | 314.8 KiB |  00m00s
[105/148] libxml2-0:2.12.9-2.fc42.x86_6 100% | 135.9 MiB/s | 696.0 KiB |  00m00s
[106/148] lz4-libs-0:1.10.0-2.fc42.x86_ 100% |  15.2 MiB/s |  78.1 KiB |  00m00s
[107/148] lua-libs-0:5.4.7-3.fc43.x86_6 100% |  31.8 MiB/s | 130.4 KiB |  00m00s
[108/148] elfutils-default-yama-scope-0 100% |   6.2 MiB/s |  12.6 KiB |  00m00s
[109/148] sqlite-libs-0:3.49.0-1.fc43.x 100% | 149.7 MiB/s | 766.3 KiB |  00m00s
[110/148] rpm-sequoia-0:1.7.0-5.fc43.x8 100% | 127.1 MiB/s | 911.1 KiB |  00m00s
[111/148] json-c-0:0.18-2.fc42.x86_64   100% |  14.6 MiB/s |  44.9 KiB |  00m00s
[112/148] libgcc-0:15.0.1-0.9.fc43.x86_ 100% |  11.6 MiB/s | 118.4 KiB |  00m00s
[113/148] libgomp-0:15.0.1-0.9.fc43.x86 100% |  28.5 MiB/s | 350.5 KiB |  00m00s
[114/148] alternatives-0:1.31-3.fc42.x8 100% |  13.3 MiB/s |  40.9 KiB |  00m00s
[115/148] libstdc++-0:15.0.1-0.9.fc43.x 100% |  45.7 MiB/s | 888.2 KiB |  00m00s
[116/148] jansson-0:2.14-2.fc42.x86_64  100% |  14.9 MiB/s |  45.7 KiB |  00m00s
[117/148] pkgconf-pkg-config-0:2.3.0-2. 100% |   4.8 MiB/s |   9.9 KiB |  00m00s
[118/148] pkgconf-0:2.3.0-2.fc42.x86_64 100% |  21.9 MiB/s |  44.9 KiB |  00m00s
[119/148] pkgconf-m4-0:2.3.0-2.fc42.noa 100% |   7.0 MiB/s |  14.2 KiB |  00m00s
[120/148] binutils-0:2.44-3.fc43.x86_64 100% | 264.2 MiB/s |   5.8 MiB |  00m00s
[121/148] libpkgconf-0:2.3.0-2.fc42.x86 100% |   4.2 MiB/s |  38.4 KiB |  00m00s
[122/148] libffi-0:3.4.7-2.fc43.x86_64  100% |   5.6 MiB/s |  40.0 KiB |  00m00s
[123/148] p11-kit-0:0.25.5-5.fc42.x86_6 100% |  96.0 MiB/s | 491.7 KiB |  00m00s
[124/148] p11-kit-trust-0:0.25.5-5.fc42 100% |  32.4 MiB/s | 132.6 KiB |  00m00s
[125/148] libtasn1-0:4.20.0-1.fc43.x86_ 100% |  14.6 MiB/s |  75.0 KiB |  00m00s
[126/148] fedora-release-0:43-0.6.noarc 100% |   7.3 MiB/s |  14.9 KiB |  00m00s
[127/148] systemd-standalone-sysusers-0 100% |  50.6 MiB/s | 155.4 KiB |  00m00s
[128/148] xxhash-libs-0:0.8.3-2.fc42.x8 100% |  12.7 MiB/s |  39.1 KiB |  00m00s
[129/148] fedora-release-identity-basic 100% |   7.7 MiB/s |  15.7 KiB |  00m00s
[130/148] libcurl-0:8.12.1-1.fc43.x86_6 100% |  74.4 MiB/s | 381.0 KiB |  00m00s
[131/148] krb5-libs-0:1.21.3-5.fc42.x86 100% | 106.7 MiB/s | 764.7 KiB |  00m00s
[132/148] libbrotli-0:1.1.0-6.fc42.x86_ 100% |  66.4 MiB/s | 339.8 KiB |  00m00s
[133/148] libidn2-0:2.3.7-3.fc42.x86_64 100% |  23.0 MiB/s | 118.0 KiB |  00m00s
[134/148] gdb-minimal-0:16.2-1.fc43.x86 100% | 184.8 MiB/s |   4.4 MiB |  00m00s
[135/148] libnghttp2-0:1.65.0-1.fc43.x8 100% |   8.9 MiB/s |  72.6 KiB |  00m00s
[136/148] libpsl-0:0.21.5-5.fc42.x86_64 100% |   8.9 MiB/s |  64.0 KiB |  00m00s
[137/148] libssh-0:0.11.1-4.fc42.x86_64 100% |  57.0 MiB/s | 233.3 KiB |  00m00s
[138/148] openldap-0:2.6.9-3.fc42.x86_6 100% |  63.5 MiB/s | 260.2 KiB |  00m00s
[139/148] keyutils-libs-0:1.6.3-5.fc42. 100% |  10.3 MiB/s |  31.5 KiB |  00m00s
[140/148] libcom_err-0:1.47.2-3.fc42.x8 100% |  13.1 MiB/s |  26.9 KiB |  00m00s
[141/148] libverto-0:0.3.2-10.fc42.x86_ 100% |  10.2 MiB/s |  20.8 KiB |  00m00s
[142/148] publicsuffix-list-dafsa-0:202 100% |  28.7 MiB/s |  58.8 KiB |  00m00s
[143/148] libunistring-0:1.1-9.fc42.x86 100% | 106.0 MiB/s | 542.5 KiB |  00m00s
[144/148] libssh-config-0:0.11.1-4.fc42 100% |   4.4 MiB/s |   9.0 KiB |  00m00s
[145/148] cyrus-sasl-lib-0:2.1.28-30.fc 100% | 155.0 MiB/s | 793.5 KiB |  00m00s
[146/148] libtool-ltdl-0:2.5.4-4.fc42.x 100% |   8.8 MiB/s |  36.2 KiB |  00m00s
[147/148] libevent-0:2.1.12-15.fc42.x86 100% |  50.8 MiB/s | 260.2 KiB |  00m00s
[148/148] gdbm-libs-1:1.23-9.fc42.x86_6 100% |  27.8 MiB/s |  57.0 KiB |  00m00s
--------------------------------------------------------------------------------
[148/148] Total                         100% | 160.3 MiB/s |  52.2 MiB |  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% | 817.0   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_64-1741555222.965739/root/var/cache/d
[  2/150] Prepare transaction           100% |   3.7 KiB/s | 148.0   B |  00m00s
[  3/150] Installing libgcc-0:15.0.1-0. 100% | 262.0 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% |   0.0   B/s |  69.2 KiB |  00m00s
[  6/150] Installing fedora-release-ide 100% |   0.0   B/s | 976.0   B |  00m00s
[  7/150] Installing fedora-repos-rawhi 100% |   0.0   B/s |   2.4 KiB |  00m00s
[  8/150] Installing fedora-gpg-keys-0: 100% |  42.7 MiB/s | 174.8 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% |  23.8 MiB/s |  24.4 KiB |  00m00s
[ 11/150] Installing fedora-release-0:4 100% |   9.3 KiB/s | 124.0   B |  00m00s
>>> Running unknown scriptlet: setup-0:2.15.0-13.fc43.noarch                    
>>> Finished unknown scriptlet: setup-0:2.15.0-13.fc43.noarch                   
>>> Scriptlet output:                                                           
>>> Creating group 'adm' with GID 4.                                            
>>> Creating group 'audio' with GID 63.                                         
>>> Creating group 'bin' with GID 1.                                            
>>> Creating group 'cdrom' with GID 11.                                         
>>> Creating group 'clock' with GID 103.                                        
>>> Creating group 'daemon' with GID 2.                                         
>>> 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.                                         
>>>                                                                             
>>> Running unknown scriptlet: setup-0:2.15.0-13.fc43.noarch                    
>>> Finished unknown scriptlet: setup-0:2.15.0-13.fc43.noarch                   
>>> Scriptlet output:                                                           
>>> Creating user 'adm' (adm) with UID 3 and GID 4.                             
>>> Creating user 'bin' (bin) with UID 1 and GID 1.                             
>>> 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 20.                       
>>> 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-0:2.15.0-13. 100% |  47.3 MiB/s | 726.7 KiB |  00m00s
>>> [RPM] /etc/hosts created as /etc/hosts.rpmnew                               
[ 13/150] Installing filesystem-0:3.18- 100% |   2.6 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% | 269.9 MiB/s | 276.4 KiB |  00m00s
[ 16/150] Installing ncurses-base-0:6.5 100% |  68.8 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% | 232.6 MiB/s | 952.8 KiB |  00m00s
[ 19/150] Installing glibc-0:2.41.9000- 100% | 180.1 MiB/s |   6.7 MiB |  00m00s
[ 20/150] Installing bash-0:5.2.37-3.fc 100% | 247.9 MiB/s |   8.2 MiB |  00m00s
[ 21/150] Installing glibc-common-0:2.4 100% |  56.7 MiB/s |   1.0 MiB |  00m00s
[ 22/150] Installing glibc-gconv-extra- 100% | 235.8 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% |   0.0   B/s |  38.4 KiB |  00m00s
[ 27/150] Installing libblkid-0:2.40.4- 100% | 257.2 MiB/s | 263.4 KiB |  00m00s
[ 28/150] Installing gmp-1:6.3.0-3.fc43 100% | 267.4 MiB/s | 821.5 KiB |  00m00s
[ 29/150] Installing popt-0:1.19-8.fc42 100% |  68.1 MiB/s | 139.4 KiB |  00m00s
[ 30/150] Installing readline-0:8.2-13. 100% | 237.8 MiB/s | 487.1 KiB |  00m00s
[ 31/150] Installing libxcrypt-0:4.4.38 100% | 280.4 MiB/s | 287.2 KiB |  00m00s
[ 32/150] Installing libzstd-0:1.5.7-1. 100% | 395.1 MiB/s | 809.1 KiB |  00m00s
[ 33/150] Installing elfutils-libelf-0: 100% | 390.1 MiB/s |   1.2 MiB |  00m00s
[ 34/150] Installing libstdc++-0:15.0.1 100% | 351.1 MiB/s |   2.8 MiB |  00m00s
[ 35/150] Installing libattr-0:2.5.2-5. 100% |   0.0   B/s |  28.1 KiB |  00m00s
[ 36/150] Installing libacl-0:2.3.2-3.f 100% |   0.0   B/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-0:4.2.1-6.fc4 100% | 271.3 MiB/s | 833.6 KiB |  00m00s
[ 39/150] Installing gawk-0:5.3.1-1.fc4 100% |  94.2 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% | 658.7 MiB/s |  11.9 MiB |  00m00s
[ 42/150] Installing file-0:5.46-1.fc42 100% |   5.0 MiB/s | 101.7 KiB |  00m00s
[ 43/150] Installing crypto-policies-0: 100% |  31.5 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% |  52.8 MiB/s |   1.0 MiB |  00m00s
[ 46/150] Installing xz-1:5.6.3-3.fc42. 100% |  68.3 MiB/s |   1.2 MiB |  00m00s
[ 47/150] Installing libeconf-0:0.7.6-1 100% |  64.7 MiB/s |  66.2 KiB |  00m00s
[ 48/150] Installing libcap-ng-0:0.8.5- 100% |  73.1 MiB/s |  74.8 KiB |  00m00s
[ 49/150] Installing audit-libs-0:4.0.3 100% | 172.6 MiB/s | 353.4 KiB |  00m00s
[ 50/150] Installing pam-libs-0:1.7.0-4 100% | 126.1 MiB/s | 129.1 KiB |  00m00s
[ 51/150] Installing libcap-0:2.73-2.fc 100% |  14.8 MiB/s | 212.1 KiB |  00m00s
[ 52/150] Installing systemd-libs-0:257 100% | 317.7 MiB/s |   2.2 MiB |  00m00s
[ 53/150] Installing libsmartcols-0:2.4 100% | 177.3 MiB/s | 181.5 KiB |  00m00s
[ 54/150] Installing libsepol-0:3.8-1.f 100% | 269.2 MiB/s | 827.0 KiB |  00m00s
[ 55/150] Installing libselinux-0:3.8-1 100% | 189.8 MiB/s | 194.3 KiB |  00m00s
[ 56/150] Installing findutils-1:4.10.0 100% |  98.6 MiB/s |   1.9 MiB |  00m00s
[ 57/150] Installing sed-0:4.9-4.fc42.x 100% |  49.7 MiB/s | 865.5 KiB |  00m00s
[ 58/150] Installing libmount-0:2.40.4- 100% | 174.5 MiB/s | 357.4 KiB |  00m00s
[ 59/150] Installing lz4-libs-0:1.10.0- 100% | 154.7 MiB/s | 158.5 KiB |  00m00s
[ 60/150] Installing lua-libs-0:5.4.7-3 100% | 271.5 MiB/s | 278.1 KiB |  00m00s
[ 61/150] Installing alternatives-0:1.3 100% |   5.1 MiB/s |  67.7 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% | 173.9 MiB/s | 178.1 KiB |  00m00s
[ 64/150] Installing p11-kit-0:0.25.5-5 100% | 104.0 MiB/s |   2.2 MiB |  00m00s
[ 65/150] Installing libunistring-0:1.1 100% | 345.3 MiB/s |   1.7 MiB |  00m00s
[ 66/150] Installing libidn2-0:2.3.7-3. 100% | 163.6 MiB/s | 335.0 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% |  18.5 MiB/s | 397.2 KiB |  00m00s
[ 69/150] Installing zstd-0:1.5.7-1.fc4 100% |  95.0 MiB/s |   1.7 MiB |  00m00s
[ 70/150] Installing util-linux-core-0: 100% |  75.1 MiB/s |   1.4 MiB |  00m00s
[ 71/150] Installing tar-2:1.35-5.fc42. 100% | 141.0 MiB/s |   3.0 MiB |  00m00s
[ 72/150] Installing libsemanage-0:3.8- 100% | 101.0 MiB/s | 310.2 KiB |  00m00s
[ 73/150] Installing shadow-utils-2:4.1 100% | 129.6 MiB/s |   4.0 MiB |  00m00s
[ 74/150] Installing systemd-standalone 100% |  20.6 MiB/s | 273.8 KiB |  00m00s
[ 75/150] Installing zip-0:3.0-43.fc42. 100% |  45.7 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 libxml2-0:2.12.9-2 100% | 100.5 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 ed-0:1.21-2.fc42.x 100% |  10.4 MiB/s | 148.8 KiB |  00m00s
[ 82/150] Installing patch-0:2.7.6-26.f 100% |  18.1 MiB/s | 260.2 KiB |  00m00s
[ 83/150] Installing filesystem-srpm-ma 100% |  38.0 MiB/s |  38.9 KiB |  00m00s
[ 84/150] Installing elfutils-default-y 100% | 408.6 KiB/s |   2.0 KiB |  00m00s
[ 85/150] Installing elfutils-libs-0:0. 100% | 220.3 MiB/s | 676.7 KiB |  00m00s
[ 86/150] Installing cpio-0:2.15-2.fc41 100% |  57.9 MiB/s |   1.1 MiB |  00m00s
[ 87/150] Installing diffutils-0:3.10-9 100% |  88.3 MiB/s |   1.6 MiB |  00m00s
[ 88/150] Installing sqlite-libs-0:3.49 100% | 304.8 MiB/s |   1.5 MiB |  00m00s
[ 89/150] Installing json-c-0:0.18-2.fc 100% |  85.9 MiB/s |  88.0 KiB |  00m00s
[ 90/150] Installing libgomp-0:15.0.1-0 100% | 262.4 MiB/s | 537.3 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% | 136.4 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 libbrotli-0:1.1.0- 100% | 274.6 MiB/s | 843.6 KiB |  00m00s
[ 97/150] Installing libnghttp2-0:1.65. 100% | 159.5 MiB/s | 163.3 KiB |  00m00s
[ 98/150] Installing keyutils-libs-0:1. 100% |  58.3 MiB/s |  59.7 KiB |  00m00s
[ 99/150] Installing libcom_err-0:1.47. 100% |   0.0   B/s |  68.2 KiB |  00m00s
[100/150] Installing libverto-0:0.3.2-1 100% |   0.0   B/s |  27.2 KiB |  00m00s
[101/150] Installing libtool-ltdl-0:2.5 100% |   0.0   B/s |  71.2 KiB |  00m00s
[102/150] Installing gdbm-libs-1:1.23-9 100% | 128.5 MiB/s | 131.6 KiB |  00m00s
[103/150] Installing cyrus-sasl-lib-0:2 100% | 128.0 MiB/s |   2.3 MiB |  00m00s
[104/150] Installing rust-srpm-macros-0 100% |   0.0   B/s |   5.6 KiB |  00m00s
[105/150] Installing qt6-srpm-macros-0: 100% |   0.0   B/s | 740.0   B |  00m00s
[106/150] Installing qt5-srpm-macros-0: 100% |   0.0   B/s | 776.0   B |  00m00s
[107/150] Installing perl-srpm-macros-0 100% |   0.0   B/s |   1.1 KiB |  00m00s
[108/150] Installing package-notes-srpm 100% |   0.0   B/s |   2.0 KiB |  00m00s
[109/150] Installing openblas-srpm-macr 100% |   0.0   B/s | 392.0   B |  00m00s
[110/150] Installing ocaml-srpm-macros- 100% |   0.0   B/s |   2.2 KiB |  00m00s
[111/150] Installing kernel-srpm-macros 100% |   0.0   B/s |   2.3 KiB |  00m00s
[112/150] Installing gnat-srpm-macros-0 100% |   0.0   B/s |   1.3 KiB |  00m00s
[113/150] Installing ghc-srpm-macros-0: 100% |   0.0   B/s |   1.0 KiB |  00m00s
[114/150] Installing fpc-srpm-macros-0: 100% |   0.0   B/s | 420.0   B |  00m00s
[115/150] Installing ansible-srpm-macro 100% |  35.4 MiB/s |  36.2 KiB |  00m00s
[116/150] Installing coreutils-common-0 100% | 384.6 MiB/s |  11.2 MiB |  00m00s
[117/150] Installing openssl-libs-1:3.2 100% | 412.4 MiB/s |   7.8 MiB |  00m00s
[118/150] Installing coreutils-0:9.6-2. 100% | 165.4 MiB/s |   5.5 MiB |  00m00s
[119/150] Installing ca-certificates-0: 100% |   2.0 MiB/s |   2.4 MiB |  00m01s
[120/150] Installing libarchive-0:3.7.7 100% | 303.6 MiB/s | 932.6 KiB |  00m00s
[121/150] Installing krb5-libs-0:1.21.3 100% | 287.5 MiB/s |   2.3 MiB |  00m00s
[122/150] Installing libssh-0:0.11.1-4. 100% | 277.1 MiB/s | 567.5 KiB |  00m00s
[123/150] Installing gzip-0:1.13-3.fc42 100% |  27.8 MiB/s | 398.4 KiB |  00m00s
[124/150] Installing rpm-sequoia-0:1.7. 100% | 402.4 MiB/s |   2.4 MiB |  00m00s
[125/150] Installing rpm-libs-0:4.20.1- 100% | 353.2 MiB/s | 723.4 KiB |  00m00s
[126/150] Installing rpm-build-libs-0:4 100% | 202.6 MiB/s | 207.4 KiB |  00m00s
[127/150] Installing libevent-0:2.1.12- 100% | 295.2 MiB/s | 906.9 KiB |  00m00s
[128/150] Installing openldap-0:2.6.9-3 100% | 214.5 MiB/s | 658.9 KiB |  00m00s
[129/150] Installing libcurl-0:8.12.1-1 100% | 277.1 MiB/s | 851.2 KiB |  00m00s
[130/150] Installing elfutils-debuginfo 100% |   6.5 MiB/s |  86.2 KiB |  00m00s
[131/150] Installing elfutils-0:0.192-8 100% | 141.4 MiB/s |   2.7 MiB |  00m00s
[132/150] Installing binutils-0:2.44-3. 100% | 319.8 MiB/s |  25.9 MiB |  00m00s
[133/150] Installing gdb-minimal-0:16.2 100% | 289.1 MiB/s |  13.3 MiB |  00m00s
[134/150] Installing debugedit-0:5.1-5. 100% |  14.7 MiB/s | 195.4 KiB |  00m00s
[135/150] Installing curl-0:8.12.1-1.fc 100% |  20.4 MiB/s | 459.7 KiB |  00m00s
[136/150] Installing rpm-0:4.20.1-1.fc4 100% |  96.1 MiB/s |   2.5 MiB |  00m00s
[137/150] Installing efi-srpm-macros-0: 100% |   0.0   B/s |  41.1 KiB |  00m00s
[138/150] Installing lua-srpm-macros-0: 100% |   0.0   B/s |   1.9 KiB |  00m00s
[139/150] Installing tree-sitter-srpm-m 100% |   0.0   B/s |   7.9 KiB |  00m00s
[140/150] Installing zig-srpm-macros-0: 100% |   0.0   B/s |   1.7 KiB |  00m00s
[141/150] Installing fonts-srpm-macros- 100% |   0.0   B/s |  57.0 KiB |  00m00s
[142/150] Installing forge-srpm-macros- 100% |   0.0   B/s |  40.3 KiB |  00m00s
[143/150] Installing go-srpm-macros-0:3 100% |   0.0   B/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% |  94.5 MiB/s | 193.5 KiB |  00m00s
[146/150] Installing rpm-build-0:4.20.1 100% |  12.4 MiB/s | 177.4 KiB |  00m00s
[147/150] Installing pyproject-srpm-mac 100% |   0.0   B/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% |  96.2 MiB/s |   3.5 MiB |  00m00s
[150/150] Installing info-0:7.2-3.fc42. 100% | 227.8 KiB/s | 358.3 KiB |  00m02s
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 3 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-1.31-3.fc42.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-8.12.1-1.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-6-2.fc42.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-43-0.6.noarch
fedora-release-common-43-0.6.noarch
fedora-release-identity-basic-43-0.6.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-3.7.7-3.fc43.x86_64
libattr-2.5.2-5.fc42.x86_64
libblkid-2.40.4-7.fc43.x86_64
libbrotli-1.1.0-6.fc42.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-8.12.1-1.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-15.0.1-0.9.fc43.x86_64
libgomp-15.0.1-0.9.fc43.x86_64
libidn2-2.3.7-3.fc42.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++-15.0.1-0.9.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
libxml2-2.12.9-2.fc42.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-4.2.1-6.fc42.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-3.2.4-2.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-1.17.0-1.fc43.noarch
python-srpm-macros-3.13-4.fc42.noarch
qt5-srpm-macros-5.15.15-1.fc42.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-2.15.0-13.fc43.noarch
shadow-utils-4.17.0-4.fc42.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_64-1741555222.965739/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-nlegjgrg/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-1741555222.965739/root.
INFO: reusing tmpfs at /var/lib/mock/fedora-rawhide-x86_64-bootstrap-1741555222.965739/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_64-1741555222.965739/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
  dnf5-5.2.11.0-1.fc43.x86_64
  dnf5-plugins-5.2.11.0-1.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% |  80.4 KiB/s |  18.9 KiB |  00m00s
 Copr repository                        100% |  95.6 KiB/s |   1.5 KiB |  00m00s
Repositories loaded.
Package "curl-8.12.1-1.fc43.x86_64" is already installed.

Package                         Arch   Version                        Repository      Size
Installing:
 cmake                          x86_64 4.0.0~rc3-3.fc43               fedora      34.4 MiB
 gcc-c++                        x86_64 15.0.1-0.9.fc43                copr_base   40.8 MiB
 git                            x86_64 2.48.1-3.fc43                  fedora      85.3 KiB
 hipblas-devel                  x86_64 6.3.0-5.fc43                   copr_base    3.1 MiB
 hipcc-libomp-devel             x86_64 18-42.rocm6.3.2.fc43           copr_base    0.0   B
 langpacks-en                   noarch 4.2-4.fc42                     fedora     400.0   B
 libcurl-devel                  x86_64 8.12.1-1.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 6.3.0-10.fc43                  copr_base    2.8 MiB
 rocm-comgr-devel               x86_64 18-42.rocm6.3.2.fc43           copr_base  103.0 KiB
 rocm-hip-devel                 x86_64 6.3.2-4.fc43                   copr_base    2.7 MiB
 rocm-rpm-macros                noarch 6.3.2-2.fc43                   fedora      18.9 KiB
 rocm-runtime-devel             x86_64 6.3.2-3.fc43                   copr_base  565.6 KiB
 wget2-wget                     x86_64 2.2.0-3.fc43                   fedora      42.0   B
 xxd                            x86_64 2:9.1.1179-1.fc43              fedora      33.3 KiB
Installing dependencies:
 abattis-cantarell-vf-fonts     noarch 0.301-14.fc42                  fedora     192.7 KiB
 annobin-docs                   noarch 12.92-1.fc43                   fedora      98.9 KiB
 annobin-plugin-gcc             x86_64 12.92-1.fc43                   fedora     993.1 KiB
 brotli                         x86_64 1.1.0-6.fc42                   fedora      31.6 KiB
 brotli-devel                   x86_64 1.1.0-6.fc42                   fedora      65.6 KiB
 clang-resource-filesystem      x86_64 20.1.0-1.fc43                  fedora      15.3 KiB
 cmake-data                     noarch 4.0.0~rc3-3.fc43               fedora       8.6 MiB
 cmake-filesystem               x86_64 4.0.0~rc3-3.fc43               fedora       0.0   B
 cmake-rpm-macros               noarch 4.0.0~rc3-3.fc43               fedora       7.6 KiB
 cpp                            x86_64 15.0.1-0.9.fc43                copr_base   37.6 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 2.6.4-2.fc42                   fedora     292.8 KiB
 fonts-filesystem               noarch 1:2.0.5-21.fc42                fedora       0.0   B
 gcc                            x86_64 15.0.1-0.9.fc43                copr_base  110.2 MiB
 gcc-plugin-annobin             x86_64 15.0.1-0.9.fc43                copr_base   57.2 KiB
 git-core                       x86_64 2.48.1-3.fc43                  fedora      22.7 MiB
 git-core-doc                   noarch 2.48.1-3.fc43                  fedora      17.4 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                  fedora     591.4 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.fc42                   copr_base   16.4 KiB
 hipcc                          x86_64 18-42.rocm6.3.2.fc43           copr_base  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 2.11.2-2.fc42                  fedora       2.9 MiB
 jsoncpp                        x86_64 1.9.6-1.fc43                   fedora     261.6 KiB
 kernel-headers                 x86_64 6.14.0-0.rc5.43.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 1.22.0-1.fc41                  fedora       5.2 MiB
 libfido2                       x86_64 1.15.0-3.fc42                  fedora     242.1 KiB
 libgcrypt                      x86_64 1.11.0-5.fc42                  fedora       1.6 MiB
 libgfortran                    x86_64 15.0.1-0.9.fc43                copr_base    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 2.3.7-3.fc42                   fedora     253.0 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 20.1.0-1.fc43                  fedora       2.2 MiB
 libomp-devel                   x86_64 20.1.0-1.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 15.0.1-0.9.fc43                copr_base  321.9 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 15.0.1-0.9.fc43                copr_base   15.9 MiB
 libtommath                     x86_64 1.3.1~rc1-5.fc42               fedora     130.4 KiB
 libusb1                        x86_64 1.0.27-8.fc42                  fedora     166.5 KiB
 libuv                          x86_64 1:1.50.0-1.fc42                fedora     566.8 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 20.1.0-1.fc43                  fedora       0.0   B
 llvm-libs                      x86_64 20.1.0-1.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 9.9p1-12.fc43                  fedora       1.4 MiB
 openssh-clients                x86_64 9.9p1-12.fc43                  fedora       2.6 MiB
 openssl-devel                  x86_64 1:3.2.4-2.fc43                 fedora       4.3 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 2.48.1-3.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 6.3.0-10.fc43                  copr_base    3.8 GiB
 rocm-clang                     x86_64 18-42.rocm6.3.2.fc43           copr_base   92.3 MiB
 rocm-clang-devel               x86_64 18-42.rocm6.3.2.fc43           copr_base   21.8 MiB
 rocm-clang-libs                x86_64 18-42.rocm6.3.2.fc43           copr_base   91.0 MiB
 rocm-clang-runtime-devel       x86_64 18-42.rocm6.3.2.fc43           copr_base    6.9 MiB
 rocm-comgr                     x86_64 18-42.rocm6.3.2.fc43           copr_base  116.3 MiB
 rocm-device-libs               x86_64 18-42.rocm6.3.2.fc43           copr_base    3.2 MiB
 rocm-hip                       x86_64 6.3.2-4.fc43                   copr_base   23.3 MiB
 rocm-libc++                    x86_64 18-42.rocm6.3.2.fc43           copr_base    1.2 MiB
 rocm-libc++-devel              x86_64 18-42.rocm6.3.2.fc43           copr_base    7.0 MiB
 rocm-lld                       x86_64 18-42.rocm6.3.2.fc43           copr_base    5.3 MiB
 rocm-llvm                      x86_64 18-42.rocm6.3.2.fc43           copr_base   68.4 MiB
 rocm-llvm-devel                x86_64 18-42.rocm6.3.2.fc43           copr_base   24.3 MiB
 rocm-llvm-filesystem           x86_64 18-42.rocm6.3.2.fc43           copr_base    0.0   B
 rocm-llvm-libs                 x86_64 18-42.rocm6.3.2.fc43           copr_base   80.7 MiB
 rocm-llvm-static               x86_64 18-42.rocm6.3.2.fc43           copr_base  234.4 MiB
 rocm-runtime                   x86_64 6.3.2-3.fc43                   copr_base    2.9 MiB
 rocsolver                      x86_64 6.3.0-5.fc43                   copr_base  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 1.17.0-5.fc43                  copr_base    2.3 MiB
 unbound-libs                   x86_64 1.22.0-14.fc43                 fedora       1.4 MiB
 vim-filesystem                 noarch 2:9.1.1179-1.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 641 MiB. Need to download 641 MiB.
After this operation, 5 GiB extra will be used (install 5 GiB, remove 0 B).
[  1/213] langpacks-en-0:4.2-4.fc42.noa 100% | 777.3 KiB/s |  10.9 KiB |  00m00s
[  2/213] git-0:2.48.1-3.fc43.x86_64    100% |   1.2 MiB/s |  51.6 KiB |  00m00s
[  3/213] wget2-wget-0:2.2.0-3.fc43.x86 100% | 336.3 KiB/s |   9.8 KiB |  00m00s
[  4/213] xxd-2:9.1.1179-1.fc43.x86_64  100% |  10.4 MiB/s |  32.1 KiB |  00m00s
[  5/213] cmake-0:4.0.0~rc3-3.fc43.x86_ 100% | 221.1 MiB/s |  12.2 MiB |  00m00s
[  6/213] gcc-c++-0:15.0.1-0.9.fc43.x86 100% | 161.2 MiB/s |  15.0 MiB |  00m00s
[  7/213] hipblas-devel-0:6.3.0-5.fc43. 100% |   2.4 MiB/s | 105.4 KiB |  00m00s
[  8/213] libcurl-devel-0:8.12.1-1.fc43 100% | 127.3 MiB/s | 912.5 KiB |  00m00s
[  9/213] hipcc-libomp-devel-0:18-42.ro 100% | 814.6 KiB/s |  13.8 KiB |  00m00s
[ 10/213] pthreadpool-devel-0:0.0^git20 100% |   1.0 MiB/s |  14.4 KiB |  00m00s
[ 11/213] rocm-comgr-devel-0:18-42.rocm 100% |   1.1 MiB/s |  32.1 KiB |  00m00s
[ 12/213] rocblas-devel-0:6.3.0-10.fc43 100% |   2.9 MiB/s | 106.3 KiB |  00m00s
[ 13/213] rocm-rpm-macros-0:6.3.2-2.fc4 100% | 392.8 KiB/s |  15.3 KiB |  00m00s
[ 14/213] rocm-hip-devel-0:6.3.2-4.fc43 100% |   3.0 MiB/s | 226.1 KiB |  00m00s
[ 15/213] git-core-0:2.48.1-3.fc43.x86_ 100% | 243.8 MiB/s |   4.9 MiB |  00m00s
[ 16/213] git-core-doc-0:2.48.1-3.fc43. 100% | 199.8 MiB/s |   3.0 MiB |  00m00s
[ 17/213] perl-File-Basename-0:2.86-515 100% |  16.8 MiB/s |  17.2 KiB |  00m00s
[ 18/213] perl-File-Find-0:1.44-515.fc4 100% |  12.4 MiB/s |  25.4 KiB |  00m00s
[ 19/213] perl-Getopt-Long-1:2.58-3.fc4 100% |  31.1 MiB/s |  63.7 KiB |  00m00s
[ 20/213] rocm-runtime-devel-0:6.3.2-3. 100% |   1.1 MiB/s |  92.8 KiB |  00m00s
[ 21/213] perl-Git-0:2.48.1-3.fc43.noar 100% |  18.7 MiB/s |  38.3 KiB |  00m00s
[ 22/213] perl-IPC-Open3-0:1.22-515.fc4 100% |   4.3 MiB/s |  21.8 KiB |  00m00s
[ 23/213] perl-PathTools-0:3.91-513.fc4 100% |  14.2 MiB/s |  87.3 KiB |  00m00s
[ 24/213] perl-TermReadKey-0:2.38-24.fc 100% |   8.6 MiB/s |  35.4 KiB |  00m00s
[ 25/213] perl-interpreter-4:5.40.1-515 100% |  23.5 MiB/s |  72.2 KiB |  00m00s
[ 26/213] perl-lib-0:0.65-515.fc42.x86_ 100% |   4.9 MiB/s |  15.0 KiB |  00m00s
[ 27/213] langpacks-core-en-0:4.2-4.fc4 100% |   1.8 MiB/s |  10.9 KiB |  00m00s
[ 28/213] hwloc-libs-0:2.11.2-2.fc42.x8 100% | 209.3 MiB/s |   2.1 MiB |  00m00s
[ 29/213] langpacks-fonts-en-0:4.2-4.fc 100% | 215.7 KiB/s |  11.2 KiB |  00m00s
[ 30/213] libpsm2-0:12.0.1-2.fc42.x86_6 100% |  28.3 MiB/s | 202.8 KiB |  00m00s
[ 31/213] openssh-clients-0:9.9p1-12.fc 100% | 120.4 MiB/s | 739.9 KiB |  00m00s
[ 32/213] libfabric-0:1.22.0-1.fc41.x86 100% |  26.0 MiB/s |   1.4 MiB |  00m00s
[ 33/213] orangefs-0:2.9.8-14.fc42.x86_ 100% |  44.0 MiB/s |   1.9 MiB |  00m00s
[ 34/213] openmpi-0:5.0.6-5.fc43.x86_64 100% |   4.5 MiB/s |   2.0 MiB |  00m00s
[ 35/213] prrte-0:3.0.6-6.fc43.x86_64   100% |   1.6 MiB/s |  55.9 KiB |  00m00s
[ 36/213] cmake-data-0:4.0.0~rc3-3.fc43 100% | 179.0 MiB/s |   2.5 MiB |  00m00s
[ 37/213] cmake-filesystem-0:4.0.0~rc3- 100% |   8.8 MiB/s |  18.0 KiB |  00m00s
[ 38/213] expat-0:2.6.4-2.fc42.x86_64   100% |  37.3 MiB/s | 114.7 KiB |  00m00s
[ 39/213] jsoncpp-0:1.9.6-1.fc43.x86_64 100% |  19.8 MiB/s | 101.6 KiB |  00m00s
[ 40/213] libuv-1:1.50.0-1.fc42.x86_64  100% |  51.7 MiB/s | 264.8 KiB |  00m00s
[ 41/213] make-1:4.4.1-10.fc42.x86_64   100% | 114.6 MiB/s | 587.0 KiB |  00m00s
[ 42/213] rhash-0:1.4.5-2.fc42.x86_64   100% |  48.5 MiB/s | 198.7 KiB |  00m00s
[ 43/213] libmpc-0:1.3.1-7.fc42.x86_64  100% |  23.1 MiB/s |  70.9 KiB |  00m00s
[ 44/213] hipcc-0:18-42.rocm6.3.2.fc43. 100% |   2.6 MiB/s | 126.6 KiB |  00m00s
[ 45/213] wget2-0:2.2.0-3.fc43.x86_64   100% |   2.1 MiB/s | 279.3 KiB |  00m00s
[ 46/213] pthreadpool-0:0.0^git20230829 100% |   2.3 MiB/s |  46.6 KiB |  00m00s
[ 47/213] pmix-0:4.2.8-4.fc42.x86_64    100% |   1.4 MiB/s | 677.0 KiB |  00m00s
[ 48/213] perl-File-Copy-0:2.41-515.fc4 100% |  19.7 MiB/s |  20.1 KiB |  00m00s
[ 49/213] perl-File-Which-0:1.27-13.fc4 100% |   5.3 MiB/s |  21.6 KiB |  00m00s
[ 50/213] perl-Getopt-Std-0:1.14-515.fc 100% |  15.3 MiB/s |  15.7 KiB |  00m00s
[ 51/213] perl-Scalar-List-Utils-5:1.68 100% |  24.3 MiB/s |  74.7 KiB |  00m00s
[ 52/213] perl-URI-0:5.31-2.fc42.noarch 100% |  45.8 MiB/s | 140.7 KiB |  00m00s
[ 53/213] rocm-hip-0:6.3.2-4.fc43.x86_6 100% |  69.8 MiB/s |   9.3 MiB |  00m00s
[ 54/213] environment-modules-0:5.5.0-3 100% |  83.0 MiB/s | 764.7 KiB |  00m00s
[ 55/213] less-0:668-2.fc42.x86_64      100% |  61.8 MiB/s | 190.0 KiB |  00m00s
[ 56/213] perl-Carp-0:1.54-512.fc42.noa 100% |  28.2 MiB/s |  28.9 KiB |  00m00s
[ 57/213] perl-Exporter-0:5.78-512.fc42 100% |  15.1 MiB/s |  31.0 KiB |  00m00s
[ 58/213] perl-Pod-Usage-4:2.03-512.fc4 100% |  19.5 MiB/s |  40.0 KiB |  00m00s
[ 59/213] perl-Text-ParseWords-0:3.31-5 100% |   8.0 MiB/s |  16.5 KiB |  00m00s
[ 60/213] perl-base-0:2.27-515.fc42.noa 100% |  15.8 MiB/s |  16.2 KiB |  00m00s
[ 61/213] perl-constant-0:1.33-513.fc42 100% |  11.2 MiB/s |  23.0 KiB |  00m00s
[ 62/213] perl-overload-0:1.37-515.fc42 100% |  22.2 MiB/s |  45.5 KiB |  00m00s
[ 63/213] perl-Error-1:0.17030-1.fc43.n 100% |  13.2 MiB/s |  40.4 KiB |  00m00s
[ 64/213] perl-Fcntl-0:1.18-515.fc42.x8 100% |   9.7 MiB/s |  29.8 KiB |  00m00s
[ 65/213] perl-IO-0:1.55-515.fc42.x86_6 100% |  39.9 MiB/s |  81.7 KiB |  00m00s
[ 66/213] perl-POSIX-0:2.20-515.fc42.x8 100% |  15.9 MiB/s |  97.7 KiB |  00m00s
[ 67/213] perl-Symbol-0:1.09-515.fc42.n 100% |  13.9 MiB/s |  14.2 KiB |  00m00s
[ 68/213] perl-Errno-0:1.38-515.fc42.x8 100% |  14.6 MiB/s |  14.9 KiB |  00m00s
[ 69/213] perl-libs-4:5.40.1-515.fc42.x 100% | 233.6 MiB/s |   2.3 MiB |  00m00s
[ 70/213] rocblas-0:6.3.0-10.fc43.x86_6 100% | 288.9 MiB/s | 192.4 MiB |  00m01s
[ 71/213] perl-DynaLoader-0:1.56-515.fc 100% | 154.1 KiB/s |  26.0 KiB |  00m00s
[ 72/213] perl-vars-0:1.05-515.fc42.noa 100% |   6.3 MiB/s |  13.0 KiB |  00m00s
[ 73/213] default-fonts-core-sans-0:4.2 100% |  15.3 MiB/s |  31.3 KiB |  00m00s
[ 74/213] google-noto-sans-mono-vf-font 100% |  16.9 MiB/s | 276.9 KiB |  00m00s
[ 75/213] libibverbs-0:56.0-2.fc43.x86_ 100% |  87.5 MiB/s | 447.9 KiB |  00m00s
[ 76/213] libnl3-0:3.11.0-3.fc42.x86_64 100% | 116.3 MiB/s | 357.2 KiB |  00m00s
[ 77/213] librdmacm-0:56.0-2.fc43.x86_6 100% |  17.5 MiB/s |  71.7 KiB |  00m00s
[ 78/213] google-noto-serif-vf-fonts-0: 100% |   5.0 MiB/s | 665.5 KiB |  00m00s
[ 79/213] libedit-0:3.1-55.20250104cvs. 100% |  34.3 MiB/s | 105.3 KiB |  00m00s
[ 80/213] libfido2-0:1.15.0-3.fc42.x86_ 100% |  24.0 MiB/s |  98.4 KiB |  00m00s
[ 81/213] openssh-0:9.9p1-12.fc43.x86_6 100% |  31.4 MiB/s | 353.6 KiB |  00m00s
[ 82/213] numactl-libs-0:2.0.19-2.fc42. 100% | 246.3 KiB/s |  31.3 KiB |  00m00s
[ 83/213] tcsh-0:6.24.14-2.fc42.x86_64  100% |  64.6 MiB/s | 462.8 KiB |  00m00s
[ 84/213] prrte-libs-0:3.0.6-6.fc43.x86 100% |  11.6 MiB/s | 545.8 KiB |  00m00s
[ 85/213] gpgme-0:1.24.2-1.fc43.x86_64  100% |  53.6 MiB/s | 219.7 KiB |  00m00s
[ 86/213] rocm-comgr-0:18-42.rocm6.3.2. 100% |  31.5 MiB/s |  29.1 MiB |  00m01s
[ 87/213] munge-libs-0:0.5.16-5.fc43.x8 100% | 196.8 KiB/s |  20.5 KiB |  00m00s
[ 88/213] emacs-filesystem-1:30.0-4.fc4 100% |   2.4 MiB/s |   7.4 KiB |  00m00s
[ 89/213] vim-filesystem-2:9.1.1179-1.f 100% |   4.0 MiB/s |  16.2 KiB |  00m00s
[ 90/213] perl-Data-Dumper-0:2.189-513. 100% |  13.8 MiB/s |  56.7 KiB |  00m00s
[ 91/213] perl-MIME-Base32-0:1.303-23.f 100% |   6.7 MiB/s |  20.5 KiB |  00m00s
[ 92/213] perl-MIME-Base64-0:3.16-512.f 100% |   9.7 MiB/s |  29.9 KiB |  00m00s
[ 93/213] perl-libnet-0:3.15-513.fc42.n 100% |  41.8 MiB/s | 128.4 KiB |  00m00s
[ 94/213] perl-parent-1:0.244-2.fc42.no 100% |   7.4 MiB/s |  15.2 KiB |  00m00s
[ 95/213] man-db-0:2.13.0-2.fc42.x86_64 100% | 131.3 MiB/s |   1.3 MiB |  00m00s
[ 96/213] perl-Pod-Perldoc-0:3.28.01-51 100% |  20.9 MiB/s |  85.8 KiB |  00m00s
[ 97/213] perl-podlators-1:6.0.2-3.fc42 100% |  31.4 MiB/s | 128.6 KiB |  00m00s
[ 98/213] perl-mro-0:1.29-515.fc42.x86_ 100% |  14.6 MiB/s |  29.9 KiB |  00m00s
[ 99/213] perl-overloading-0:0.02-515.f 100% |   6.3 MiB/s |  12.9 KiB |  00m00s
[100/213] perl-File-stat-0:1.14-515.fc4 100% |  16.7 MiB/s |  17.1 KiB |  00m00s
[101/213] perl-SelectSaver-0:1.02-515.f 100% |  11.5 MiB/s |  11.7 KiB |  00m00s
[102/213] perl-Socket-4:2.038-512.fc42. 100% |  17.8 MiB/s |  54.8 KiB |  00m00s
[103/213] perl-locale-0:1.12-515.fc42.n 100% |   6.7 MiB/s |  13.6 KiB |  00m00s
[104/213] abattis-cantarell-vf-fonts-0: 100% |  58.7 MiB/s | 120.3 KiB |  00m00s
[105/213] google-noto-sans-vf-fonts-0:2 100% | 100.0 MiB/s | 614.5 KiB |  00m00s
[106/213] fonts-filesystem-1:2.0.5-21.f 100% |   4.2 MiB/s |   8.6 KiB |  00m00s
[107/213] google-noto-fonts-common-0:20 100% |   8.3 MiB/s |  17.1 KiB |  00m00s
[108/213] libcbor-0:0.11.0-3.fc42.x86_6 100% |  16.2 MiB/s |  33.3 KiB |  00m00s
[109/213] rocm-device-libs-0:18-42.rocm 100% |   6.6 MiB/s | 489.1 KiB |  00m00s
[110/213] libassuan-0:2.5.7-3.fc42.x86_ 100% |  22.0 MiB/s |  67.6 KiB |  00m00s
[111/213] wget2-libs-0:2.2.0-3.fc43.x86 100% |   1.1 MiB/s | 146.4 KiB |  00m00s
[112/213] libgpg-error-0:1.51-2.fc42.x8 100% |  57.9 MiB/s | 237.2 KiB |  00m00s
[113/213] gnupg2-0:2.4.7-2.fc42.x86_64  100% | 185.1 MiB/s |   2.8 MiB |  00m00s
[114/213] gnutls-0:3.8.9-5.fc43.x86_64  100% | 112.1 MiB/s |   1.2 MiB |  00m00s
[115/213] gnutls-dane-0:3.8.9-5.fc43.x8 100% |   5.9 MiB/s |  42.3 KiB |  00m00s
[116/213] perl-Digest-MD5-0:2.59-6.fc42 100% |  17.6 MiB/s |  36.0 KiB |  00m00s
[117/213] perl-FileHandle-0:2.05-515.fc 100% |   7.6 MiB/s |  15.5 KiB |  00m00s
[118/213] perl-B-0:1.89-515.fc42.x86_64 100% |  43.2 MiB/s | 177.0 KiB |  00m00s
[119/213] perl-IO-Socket-IP-0:0.43-2.fc 100% |  20.7 MiB/s |  42.4 KiB |  00m00s
[120/213] perl-Time-Local-2:1.350-512.f 100% |  11.2 MiB/s |  34.5 KiB |  00m00s
[121/213] libpipeline-0:1.5.8-2.fc42.x8 100% |  29.3 MiB/s |  60.0 KiB |  00m00s
[122/213] groff-base-0:1.23.0-8.fc42.x8 100% | 157.8 MiB/s |   1.1 MiB |  00m00s
[123/213] perl-File-Temp-1:0.231.100-51 100% |  11.6 MiB/s |  59.2 KiB |  00m00s
[124/213] perl-HTTP-Tiny-0:0.090-2.fc42 100% |  18.4 MiB/s |  56.5 KiB |  00m00s
[125/213] perl-Term-Cap-0:1.18-512.fc42 100% |  10.8 MiB/s |  22.2 KiB |  00m00s
[126/213] perl-Pod-Simple-1:3.45-512.fc 100% |  53.5 MiB/s | 219.0 KiB |  00m00s
[127/213] perl-Term-ANSIColor-0:5.01-51 100% |  11.6 MiB/s |  47.7 KiB |  00m00s
[128/213] perl-Class-Struct-0:0.68-515. 100% |  10.8 MiB/s |  22.1 KiB |  00m00s
[129/213] libksba-0:1.6.7-3.fc42.x86_64 100% |  52.7 MiB/s | 162.0 KiB |  00m00s
[130/213] npth-0:1.8-2.fc42.x86_64      100% |  12.6 MiB/s |  25.8 KiB |  00m00s
[131/213] nettle-0:3.10.1-1.fc43.x86_64 100% |  82.9 MiB/s | 424.6 KiB |  00m00s
[132/213] tpm2-tss-0:4.1.3-6.fc42.x86_6 100% |  69.2 MiB/s | 425.4 KiB |  00m00s
[133/213] perl-if-0:0.61.000-515.fc42.n 100% |   3.4 MiB/s |  14.0 KiB |  00m00s
[134/213] unbound-libs-0:1.22.0-14.fc43 100% |  91.0 MiB/s | 559.1 KiB |  00m00s
[135/213] libgcrypt-0:1.11.0-5.fc42.x86 100% |  36.2 MiB/s | 593.3 KiB |  00m00s
[136/213] perl-Digest-0:1.20-512.fc42.n 100% |   8.1 MiB/s |  24.9 KiB |  00m00s
[137/213] perl-File-Path-0:2.18-512.fc4 100% |  17.2 MiB/s |  35.2 KiB |  00m00s
[138/213] perl-IO-Socket-SSL-0:2.089-2. 100% |  45.0 MiB/s | 230.2 KiB |  00m00s
[139/213] perl-Net-SSLeay-0:1.94-8.fc42 100% |  91.8 MiB/s | 376.0 KiB |  00m00s
[140/213] perl-Pod-Escapes-1:1.07-512.f 100% |   3.2 MiB/s |  19.8 KiB |  00m00s
[141/213] perl-Text-Tabs+Wrap-0:2024.00 100% |   7.1 MiB/s |  21.8 KiB |  00m00s
[142/213] libusb1-0:1.0.27-8.fc42.x86_6 100% |  25.2 MiB/s |  77.4 KiB |  00m00s
[143/213] ncurses-0:6.5-5.20250125.fc42 100% |  69.1 MiB/s | 424.5 KiB |  00m00s
[144/213] hiredis-0:1.2.0-6.fc42.x86_64 100% |  12.4 MiB/s |  50.7 KiB |  00m00s
[145/213] protobuf-c-0:1.5.0-4.fc41.x86 100% |   7.9 MiB/s |  32.4 KiB |  00m00s
[146/213] perl-AutoLoader-0:5.74-515.fc 100% |  10.4 MiB/s |  21.2 KiB |  00m00s
[147/213] libb2-0:0.98.1-13.fc42.x86_64 100% |   5.0 MiB/s |  25.4 KiB |  00m00s
[148/213] mpdecimal-0:4.0.0-2.fc43.x86_ 100% |  15.8 MiB/s |  97.0 KiB |  00m00s
[149/213] tzdata-0:2025a-1.fc43.noarch  100% |  99.5 MiB/s | 713.4 KiB |  00m00s
[150/213] python-pip-wheel-0:24.3.1-2.f 100% | 120.4 MiB/s |   1.2 MiB |  00m00s
[151/213] libdrm-0:2.4.124-2.fc42.x86_6 100% |   8.7 MiB/s | 161.0 KiB |  00m00s
[152/213] libpciaccess-0:0.16-15.fc42.x 100% |   2.1 MiB/s |  26.3 KiB |  00m00s
[153/213] python3-libs-0:3.13.2-2.fc43. 100% | 148.1 MiB/s |   9.2 MiB |  00m00s
[154/213] hwdata-0:0.393-1.fc43.noarch  100% |  86.1 MiB/s |   1.6 MiB |  00m00s
[155/213] rocm-runtime-0:6.3.2-3.fc43.x 100% |   8.6 MiB/s | 615.9 KiB |  00m00s
[156/213] rocm-clang-devel-0:18-42.rocm 100% |  16.7 MiB/s |   2.3 MiB |  00m00s
[157/213] rocm-clang-libs-0:18-42.rocm6 100% | 103.6 MiB/s |  21.4 MiB |  00m00s
[158/213] rocm-clang-runtime-devel-0:18 100% |   4.5 MiB/s | 487.6 KiB |  00m00s
[159/213] rocm-clang-0:18-42.rocm6.3.2. 100% |  70.2 MiB/s |  20.9 MiB |  00m00s
[160/213] rocm-libc++-devel-0:18-42.roc 100% |  10.3 MiB/s | 833.4 KiB |  00m00s
[161/213] rocm-libc++-0:18-42.rocm6.3.2 100% |   5.9 MiB/s | 342.5 KiB |  00m00s
[162/213] rocm-llvm-filesystem-0:18-42. 100% | 422.0 KiB/s |  21.9 KiB |  00m00s
[163/213] rocm-lld-0:18-42.rocm6.3.2.fc 100% |  17.2 MiB/s |   1.4 MiB |  00m00s
[164/213] rocm-llvm-devel-0:18-42.rocm6 100% |  24.8 MiB/s |   3.6 MiB |  00m00s
[165/213] rocm-llvm-libs-0:18-42.rocm6. 100% |  53.1 MiB/s |  19.3 MiB |  00m00s
[166/213] python3-0:3.13.2-2.fc43.x86_6 100% |  13.9 MiB/s |  28.4 KiB |  00m00s
[167/213] libomp-devel-0:20.1.0-1.fc43. 100% |  30.5 MiB/s | 281.3 KiB |  00m00s
[168/213] clang-resource-filesystem-0:2 100% |   6.6 MiB/s |  20.2 KiB |  00m00s
[169/213] libomp-0:20.1.0-1.fc43.x86_64 100% |  89.9 MiB/s | 736.8 KiB |  00m00s
[170/213] rocm-llvm-static-0:18-42.rocm 100% |  70.4 MiB/s |  27.5 MiB |  00m00s
[171/213] llvm-libs-0:20.1.0-1.fc43.x86 100% | 217.3 MiB/s |  33.5 MiB |  00m00s
[172/213] llvm-filesystem-0:20.1.0-1.fc 100% | 433.5 KiB/s |  14.3 KiB |  00m00s
[173/213] hipblas-0:6.3.0-5.fc43.x86_64 100% |   7.5 MiB/s | 160.8 KiB |  00m00s
[174/213] hipblas-common-devel-0:6.3.0- 100% | 268.0 KiB/s |  13.4 KiB |  00m00s
[175/213] rocm-llvm-0:18-42.rocm6.3.2.f 100% |  52.3 MiB/s |  15.8 MiB |  00m00s
[176/213] libstdc++-devel-0:15.0.1-0.9. 100% |  67.0 MiB/s |   2.7 MiB |  00m00s
[177/213] cpp-0:15.0.1-0.9.fc43.x86_64  100% | 103.5 MiB/s |  12.7 MiB |  00m00s
[178/213] glibc-devel-0:2.41.9000-2.fc4 100% | 127.7 MiB/s | 653.7 KiB |  00m00s
[179/213] gcc-0:15.0.1-0.9.fc43.x86_64  100% | 148.8 MiB/s |  38.7 MiB |  00m00s
[180/213] libxcrypt-devel-0:4.4.38-6.fc 100% | 813.3 KiB/s |  29.3 KiB |  00m00s
[181/213] perl-Encode-4:3.21-512.fc42.x 100% | 175.4 MiB/s |   1.1 MiB |  00m00s
[182/213] perl-Storable-1:3.32-512.fc42 100% |  13.9 MiB/s |  99.6 KiB |  00m00s
[183/213] libquadmath-0:15.0.1-0.9.fc43 100% |   5.6 MiB/s | 188.6 KiB |  00m00s
[184/213] libgfortran-0:15.0.1-0.9.fc43 100% |  21.9 MiB/s | 941.4 KiB |  00m00s
[185/213] brotli-devel-0:1.1.0-6.fc42.x 100% |  16.6 MiB/s |  33.9 KiB |  00m00s
[186/213] brotli-0:1.1.0-6.fc42.x86_64  100% |   6.5 MiB/s |  19.9 KiB |  00m00s
[187/213] krb5-devel-0:1.21.3-5.fc42.x8 100% |  34.8 MiB/s | 142.5 KiB |  00m00s
[188/213] libkadm5-0:1.21.3-5.fc42.x86_ 100% |  25.2 MiB/s |  77.4 KiB |  00m00s
[189/213] libidn2-devel-0:2.3.7-3.fc42. 100% |  23.0 MiB/s |  70.7 KiB |  00m00s
[190/213] libnghttp2-devel-0:1.65.0-1.f 100% |  26.7 MiB/s |  54.7 KiB |  00m00s
[191/213] libpsl-devel-0:0.21.5-5.fc42. 100% |  16.2 MiB/s |  33.2 KiB |  00m00s
[192/213] publicsuffix-list-0:20250116- 100% |  21.7 MiB/s |  89.0 KiB |  00m00s
[193/213] libssh-devel-0:0.11.1-4.fc42. 100% |  20.4 MiB/s |  41.8 KiB |  00m00s
[194/213] openssl-devel-1:3.2.4-2.fc43. 100% | 201.0 MiB/s |   2.8 MiB |  00m00s
[195/213] zlib-ng-compat-devel-0:2.2.4- 100% |  12.5 MiB/s |  38.3 KiB |  00m00s
[196/213] keyutils-libs-devel-0:1.6.3-5 100% |  19.5 MiB/s |  59.9 KiB |  00m00s
[197/213] libcom_err-devel-0:1.47.2-3.f 100% |   8.2 MiB/s |  16.7 KiB |  00m00s
[198/213] libselinux-devel-0:3.8-1.fc42 100% |  37.0 MiB/s | 151.7 KiB |  00m00s
[199/213] libsepol-devel-0:3.8-1.fc42.x 100% |  23.6 MiB/s |  48.4 KiB |  00m00s
[200/213] libverto-devel-0:0.3.2-10.fc4 100% |   7.0 MiB/s |  14.4 KiB |  00m00s
[201/213] kernel-headers-0:6.14.0-0.rc5 100% | 165.3 MiB/s |   1.7 MiB |  00m00s
[202/213] procps-ng-0:4.0.4-6.fc42.x86_ 100% |  71.4 MiB/s | 365.3 KiB |  00m00s
[203/213] tcl-1:9.0.0-8.fc43.x86_64     100% |  95.2 MiB/s |   1.2 MiB |  00m00s
[204/213] libtommath-0:1.3.1~rc1-5.fc42 100% |  21.0 MiB/s |  64.4 KiB |  00m00s
[205/213] pcre2-devel-0:10.45-1.fc43.x8 100% |  75.8 MiB/s | 543.4 KiB |  00m00s
[206/213] rocsolver-0:6.3.0-5.fc43.x86_ 100% | 201.0 MiB/s | 109.9 MiB |  00m01s
[207/213] pcre2-utf16-0:10.45-1.fc43.x8 100% |   2.3 MiB/s | 241.9 KiB |  00m00s
[208/213] ucx-0:1.17.0-5.fc43.x86_64    100% |   3.7 MiB/s | 823.6 KiB |  00m00s
[209/213] pcre2-utf32-0:10.45-1.fc43.x8 100% |  74.5 MiB/s | 228.8 KiB |  00m00s
[210/213] cmake-rpm-macros-0:4.0.0~rc3- 100% |   4.3 MiB/s |  17.4 KiB |  00m00s
[211/213] annobin-plugin-gcc-0:12.92-1. 100% | 120.1 MiB/s | 983.7 KiB |  00m00s
[212/213] annobin-docs-0:12.92-1.fc43.n 100% |  12.9 MiB/s |  92.5 KiB |  00m00s
[213/213] gcc-plugin-annobin-0:15.0.1-0 100% |   8.4 MiB/s |  43.0 KiB |  00m00s
--------------------------------------------------------------------------------
[213/213] Total                         100% | 186.8 MiB/s | 640.9 MiB |  00m03s
Running transaction
[  1/215] Verify package files          100% |  97.0   B/s | 213.0   B |  00m02s
[  2/215] Prepare transaction           100% |   1.5 KiB/s | 213.0   B |  00m00s
[  3/215] Installing cmake-filesystem-0 100% |   7.3 MiB/s |   7.5 KiB |  00m00s
[  4/215] Installing libgpg-error-0:1.5 100% |  54.9 MiB/s | 900.0 KiB |  00m00s
[  5/215] Installing fonts-filesystem-1 100% |   0.0   B/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-0:2.11. 100% | 477.7 MiB/s |   2.9 MiB |  00m00s
[  8/215] Installing google-noto-fonts- 100% |   0.0   B/s |  18.5 KiB |  00m00s
[  9/215] Installing libnl3-0:3.11.0-3. 100% | 255.9 MiB/s |   1.0 MiB |  00m00s
[ 10/215] Installing libibverbs-0:56.0- 100% | 233.8 MiB/s |   1.2 MiB |  00m00s
[ 11/215] Installing less-0:668-2.fc42. 100% |  28.5 MiB/s | 409.1 KiB |  00m00s
[ 12/215] Installing libmpc-0:1.3.1-7.f 100% | 162.2 MiB/s | 166.1 KiB |  00m00s
[ 13/215] Installing expat-0:2.6.4-2.fc 100% |  22.2 MiB/s | 294.9 KiB |  00m00s
[ 14/215] Installing libpsm2-0:12.0.1-2 100% | 216.5 MiB/s | 443.4 KiB |  00m00s
[ 15/215] Installing libassuan-0:2.5.7- 100% | 165.6 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 rocm-llvm-filesyst 100% |   7.0 MiB/s |  14.3 KiB |  00m00s
[ 18/215] Installing rocm-libc++-0:18-4 100% |  63.7 MiB/s |   1.2 MiB |  00m00s
[ 19/215] Installing rocm-llvm-libs-0:1 100% |  73.8 MiB/s |  80.7 MiB |  00m01s
[ 20/215] Installing rocm-clang-libs-0: 100% |  75.4 MiB/s |  91.0 MiB |  00m01s
[ 21/215] Installing nettle-0:3.10.1-1. 100% | 258.3 MiB/s | 793.6 KiB |  00m00s
[ 22/215] Installing gnutls-0:3.8.9-5.f 100% | 223.3 MiB/s |   3.6 MiB |  00m00s
[ 23/215] Installing groff-base-0:1.23. 100% | 114.5 MiB/s |   3.9 MiB |  00m00s
[ 24/215] Installing vim-filesystem-2:9 100% |   4.6 MiB/s |   4.7 KiB |  00m00s
[ 25/215] Installing libedit-0:3.1-55.2 100% | 240.0 MiB/s | 245.8 KiB |  00m00s
[ 26/215] Installing rocm-comgr-0:18-42 100% |  71.6 MiB/s | 116.3 MiB |  00m02s
[ 27/215] Installing make-1:4.4.1-10.fc 100% |  94.7 MiB/s |   1.8 MiB |  00m00s
[ 28/215] Installing rocm-lld-0:18-42.r 100% |  67.4 MiB/s |   5.3 MiB |  00m00s
[ 29/215] Installing rocm-libc++-devel- 100% |  94.9 MiB/s |   7.2 MiB |  00m00s
[ 30/215] Installing cpp-0:15.0.1-0.9.f 100% | 327.0 MiB/s |  37.6 MiB |  00m00s
[ 31/215] Installing librdmacm-0:56.0-2 100% | 140.4 MiB/s | 143.8 KiB |  00m00s
[ 32/215] Installing libfabric-0:1.22.0 100% | 215.2 MiB/s |   5.2 MiB |  00m00s
[ 33/215] Installing google-noto-sans-m 100% | 274.5 MiB/s | 562.2 KiB |  00m00s
[ 34/215] Installing google-noto-serif- 100% | 318.0 MiB/s |   1.6 MiB |  00m00s
[ 35/215] Installing google-noto-sans-v 100% | 347.8 MiB/s |   1.4 MiB |  00m00s
[ 36/215] Installing abattis-cantarell- 100% | 189.9 MiB/s | 194.4 KiB |  00m00s
[ 37/215] Installing default-fonts-core 100% |  17.8 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% | 392.3 MiB/s |   1.6 MiB |  00m00s
[ 41/215] Installing libksba-0:1.6.7-3. 100% | 395.6 MiB/s | 405.1 KiB |  00m00s
[ 42/215] Installing hipblas-common-dev 100% |   0.0   B/s |  17.8 KiB |  00m00s
[ 43/215] Installing libssh-devel-0:0.1 100% | 176.3 MiB/s | 180.5 KiB |  00m00s
[ 44/215] Installing annobin-docs-0:12. 100% |   0.0   B/s | 100.0 KiB |  00m00s
[ 45/215] Installing pcre2-utf32-0:10.4 100% | 292.5 MiB/s | 599.1 KiB |  00m00s
[ 46/215] Installing pcre2-utf16-0:10.4 100% | 306.2 MiB/s | 627.2 KiB |  00m00s
[ 47/215] Installing pcre2-devel-0:10.4 100% | 104.6 MiB/s |   2.1 MiB |  00m00s
[ 48/215] Installing libtommath-0:1.3.1 100% | 128.4 MiB/s | 131.5 KiB |  00m00s
[ 49/215] Installing tcl-1:9.0.0-8.fc43 100% | 160.5 MiB/s |   4.3 MiB |  00m00s
[ 50/215] Installing procps-ng-0:4.0.4- 100% |  56.1 MiB/s |   1.0 MiB |  00m00s
[ 51/215] Installing kernel-headers-0:6 100% | 208.6 MiB/s |   6.7 MiB |  00m00s
[ 52/215] Installing libxcrypt-devel-0: 100% |  16.2 MiB/s |  33.1 KiB |  00m00s
[ 53/215] Installing glibc-devel-0:2.41 100% | 166.7 MiB/s |   2.3 MiB |  00m00s
[ 54/215] Installing gcc-0:15.0.1-0.9.f 100% | 395.1 MiB/s | 110.2 MiB |  00m00s
[ 55/215] Installing libverto-devel-0:0 100% |  25.7 MiB/s |  26.4 KiB |  00m00s
[ 56/215] Installing libsepol-devel-0:3 100% |  62.6 MiB/s | 128.3 KiB |  00m00s
[ 57/215] Installing libselinux-devel-0 100% |  39.5 MiB/s | 161.6 KiB |  00m00s
[ 58/215] Installing libcom_err-devel-0 100% |   1.4 MiB/s |  18.3 KiB |  00m00s
[ 59/215] Installing keyutils-libs-deve 100% |   7.7 MiB/s |  55.2 KiB |  00m00s
[ 60/215] Installing openssl-devel-1:3. 100% |  65.7 MiB/s |   5.2 MiB |  00m00s
[ 61/215] Installing publicsuffix-list- 100% | 161.5 MiB/s | 330.8 KiB |  00m00s
[ 62/215] Installing libpsl-devel-0:0.2 100% | 110.9 MiB/s | 113.6 KiB |  00m00s
[ 63/215] Installing libnghttp2-devel-0 100% | 280.7 MiB/s | 287.5 KiB |  00m00s
[ 64/215] Installing libidn2-devel-0:2. 100% | 127.4 MiB/s | 260.9 KiB |  00m00s
[ 65/215] Installing libkadm5-0:1.21.3- 100% | 210.8 MiB/s | 215.9 KiB |  00m00s
[ 66/215] Installing krb5-devel-0:1.21. 100% |  46.6 MiB/s | 715.2 KiB |  00m00s
[ 67/215] Installing brotli-0:1.1.0-6.f 100% |   2.6 MiB/s |  32.3 KiB |  00m00s
[ 68/215] Installing brotli-devel-0:1.1 100% |  66.4 MiB/s |  68.0 KiB |  00m00s
[ 69/215] Installing ucx-0:1.17.0-5.fc4 100% | 124.0 MiB/s |   2.4 MiB |  00m00s
[ 70/215] Installing libquadmath-0:15.0 100% | 315.6 MiB/s | 323.2 KiB |  00m00s
[ 71/215] Installing libgfortran-0:15.0 100% | 365.8 MiB/s |   3.3 MiB |  00m00s
[ 72/215] Installing libstdc++-devel-0: 100% | 381.2 MiB/s |  16.0 MiB |  00m00s
[ 73/215] Installing llvm-filesystem-0: 100% |   0.0   B/s |   1.1 KiB |  00m00s
[ 74/215] Installing llvm-libs-0:20.1.0 100% | 450.6 MiB/s | 137.0 MiB |  00m00s
[ 75/215] Installing libomp-0:20.1.0-1. 100% | 369.6 MiB/s |   2.2 MiB |  00m00s
[ 76/215] Installing clang-resource-fil 100% |   0.0   B/s |  16.7 KiB |  00m00s
[ 77/215] Installing libomp-devel-0:20. 100% | 386.6 MiB/s |   1.5 MiB |  00m00s
[ 78/215] Installing rocm-clang-runtime 100% | 138.2 MiB/s |   6.9 MiB |  00m00s
[ 79/215] Installing hwdata-0:0.393-1.f 100% | 496.4 MiB/s |   9.4 MiB |  00m00s
[ 80/215] Installing libpciaccess-0:0.1 100% |  44.8 MiB/s |  45.9 KiB |  00m00s
[ 81/215] Installing libdrm-0:2.4.124-2 100% | 201.1 MiB/s | 411.8 KiB |  00m00s
[ 82/215] Installing rocm-runtime-0:6.3 100% | 485.1 MiB/s |   2.9 MiB |  00m00s
[ 83/215] Installing rocm-runtime-devel 100% | 111.2 MiB/s | 569.2 KiB |  00m00s
[ 84/215] Installing tzdata-0:2025a-1.f 100% |  60.8 MiB/s |   1.9 MiB |  00m00s
[ 85/215] Installing python-pip-wheel-0 100% | 622.1 MiB/s |   1.2 MiB |  00m00s
[ 86/215] Installing mpdecimal-0:4.0.0- 100% | 213.2 MiB/s | 218.4 KiB |  00m00s
[ 87/215] Installing libb2-0:0.98.1-13. 100% |   7.7 MiB/s |  47.2 KiB |  00m00s
[ 88/215] Installing python3-libs-0:3.1 100% | 324.9 MiB/s |  40.3 MiB |  00m00s
[ 89/215] Installing python3-0:3.13.2-2 100% |   2.0 MiB/s |  29.4 KiB |  00m00s
[ 90/215] Installing cmake-rpm-macros-0 100% |   8.1 MiB/s |   8.3 KiB |  00m00s
[ 91/215] Installing rocm-llvm-0:18-42. 100% |  74.2 MiB/s |  68.4 MiB |  00m01s
[ 92/215] Installing rocm-llvm-devel-0: 100% |  94.2 MiB/s |  24.7 MiB |  00m00s
[ 93/215] Installing rocm-llvm-static-0 100% | 104.8 MiB/s | 234.4 MiB |  00m02s
[ 94/215] Installing protobuf-c-0:1.5.0 100% |  54.2 MiB/s |  55.5 KiB |  00m00s
[ 95/215] Installing hiredis-0:1.2.0-6. 100% |   8.1 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.    
>>>                                                                             
[ 96/215] Installing unbound-libs-0:1.2 100% | 287.9 MiB/s |   1.4 MiB |  00m00s
[ 97/215] Installing gnutls-dane-0:3.8. 100% |  68.5 MiB/s |  70.2 KiB |  00m00s
[ 98/215] Installing libusb1-0:1.0.27-8 100% |  13.7 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.   
>>>                                                                             
[ 99/215] Installing tpm2-tss-0:4.1.3-6 100% | 261.3 MiB/s |   1.6 MiB |  00m00s
[100/215] Installing ncurses-0:6.5-5.20 100% |  35.3 MiB/s | 614.7 KiB |  00m00s
[101/215] Installing perl-Digest-0:1.20 100% |  36.2 MiB/s |  37.1 KiB |  00m00s
[102/215] Installing perl-FileHandle-0: 100% |   0.0   B/s |   9.8 KiB |  00m00s
[103/215] Installing perl-B-0:1.89-515. 100% | 244.8 MiB/s | 501.3 KiB |  00m00s
[104/215] Installing perl-Digest-MD5-0: 100% |  60.1 MiB/s |  61.6 KiB |  00m00s
[105/215] Installing perl-MIME-Base32-0 100% |   0.0   B/s |  32.2 KiB |  00m00s
[106/215] Installing perl-Data-Dumper-0 100% | 114.7 MiB/s | 117.5 KiB |  00m00s
[107/215] Installing perl-libnet-0:3.15 100% | 143.9 MiB/s | 294.7 KiB |  00m00s
[108/215] Installing perl-IO-Socket-IP- 100% |  99.8 MiB/s | 102.2 KiB |  00m00s
[109/215] Installing perl-URI-0:5.31-2. 100% |  87.8 MiB/s | 269.6 KiB |  00m00s
[110/215] Installing perl-AutoLoader-0: 100% |   0.0   B/s |  20.9 KiB |  00m00s
[111/215] Installing perl-locale-0:1.12 100% |   0.0   B/s |   6.9 KiB |  00m00s
[112/215] Installing perl-Time-Local-2: 100% |  68.9 MiB/s |  70.6 KiB |  00m00s
[113/215] Installing perl-if-0:0.61.000 100% |   0.0   B/s |   6.2 KiB |  00m00s
[114/215] Installing perl-File-Path-0:2 100% |   0.0   B/s |  64.5 KiB |  00m00s
[115/215] Installing perl-Pod-Escapes-1 100% |   0.0   B/s |  25.9 KiB |  00m00s
[116/215] Installing perl-Text-Tabs+Wra 100% |   0.0   B/s |  23.9 KiB |  00m00s
[117/215] Installing perl-IO-Socket-SSL 100% | 230.3 MiB/s | 707.4 KiB |  00m00s
[118/215] Installing perl-Net-SSLeay-0: 100% | 271.7 MiB/s |   1.4 MiB |  00m00s
[119/215] Installing perl-POSIX-0:2.20- 100% | 226.7 MiB/s | 232.2 KiB |  00m00s
[120/215] Installing perl-Term-ANSIColo 100% |  96.9 MiB/s |  99.2 KiB |  00m00s
[121/215] Installing perl-Term-Cap-0:1. 100% |   0.0   B/s |  30.6 KiB |  00m00s
[122/215] Installing perl-IPC-Open3-0:1 100% |   0.0   B/s |  23.3 KiB |  00m00s
[123/215] Installing perl-Class-Struct- 100% |   0.0   B/s |  25.9 KiB |  00m00s
[124/215] Installing perl-File-Temp-1:0 100% | 160.2 MiB/s | 164.1 KiB |  00m00s
[125/215] Installing perl-HTTP-Tiny-0:0 100% | 152.8 MiB/s | 156.4 KiB |  00m00s
[126/215] Installing perl-Pod-Simple-1: 100% | 278.5 MiB/s | 570.4 KiB |  00m00s
[127/215] Installing perl-Symbol-0:1.09 100% |   0.0   B/s |   7.2 KiB |  00m00s
[128/215] Installing perl-SelectSaver-0 100% |   0.0   B/s |   2.6 KiB |  00m00s
[129/215] Installing perl-Socket-4:2.03 100% | 119.1 MiB/s | 122.0 KiB |  00m00s
[130/215] Installing perl-File-stat-0:1 100% |   0.0   B/s |  13.1 KiB |  00m00s
[131/215] Installing perl-Pod-Perldoc-0 100% |  11.0 MiB/s | 169.2 KiB |  00m00s
[132/215] Installing perl-podlators-1:6 100% |  22.4 MiB/s | 321.4 KiB |  00m00s
[133/215] Installing perl-Text-ParseWor 100% |   0.0   B/s |  14.6 KiB |  00m00s
[134/215] Installing perl-base-0:2.27-5 100% |   0.0   B/s |  12.9 KiB |  00m00s
[135/215] Installing perl-Fcntl-0:1.18- 100% |  48.9 MiB/s |  50.0 KiB |  00m00s
[136/215] Installing perl-mro-0:1.29-51 100% |   0.0   B/s |  42.6 KiB |  00m00s
[137/215] Installing perl-overloading-0 100% |   0.0   B/s |   5.5 KiB |  00m00s
[138/215] Installing perl-IO-0:1.55-515 100% | 147.6 MiB/s | 151.1 KiB |  00m00s
[139/215] Installing perl-Pod-Usage-4:2 100% |   6.5 MiB/s |  86.3 KiB |  00m00s
[140/215] Installing perl-Getopt-Std-0: 100% |   0.0   B/s |  11.7 KiB |  00m00s
[141/215] Installing perl-Scalar-List-U 100% | 145.1 MiB/s | 148.5 KiB |  00m00s
[142/215] Installing perl-constant-0:1. 100% |   0.0   B/s |  27.4 KiB |  00m00s
[143/215] Installing perl-Errno-0:1.38- 100% |   0.0   B/s |   8.7 KiB |  00m00s
[144/215] Installing perl-vars-0:1.05-5 100% |   0.0   B/s |   4.3 KiB |  00m00s
[145/215] Installing perl-MIME-Base64-0 100% |  43.2 MiB/s |  44.3 KiB |  00m00s
[146/215] Installing perl-parent-1:0.24 100% |   0.0   B/s |  11.0 KiB |  00m00s
[147/215] Installing perl-overload-0:1. 100% |   0.0   B/s |  71.9 KiB |  00m00s
[148/215] Installing perl-Storable-1:3. 100% | 228.4 MiB/s | 233.9 KiB |  00m00s
[149/215] Installing perl-Getopt-Long-1 100% | 143.8 MiB/s | 147.2 KiB |  00m00s
[150/215] Installing perl-File-Basename 100% |   0.0   B/s |  14.6 KiB |  00m00s
[151/215] Installing perl-Carp-0:1.54-5 100% |   0.0   B/s |  47.7 KiB |  00m00s
[152/215] Installing perl-Exporter-0:5. 100% |  10.9 MiB/s |  55.6 KiB |  00m00s
[153/215] Installing perl-PathTools-0:3 100% | 180.2 MiB/s | 184.5 KiB |  00m00s
[154/215] Installing perl-DynaLoader-0: 100% |   0.0   B/s |  32.5 KiB |  00m00s
[155/215] Installing perl-Encode-4:3.21 100% | 180.5 MiB/s |   4.7 MiB |  00m00s
[156/215] Installing perl-libs-4:5.40.1 100% | 267.7 MiB/s |   9.9 MiB |  00m00s
[157/215] Installing perl-interpreter-4 100% |   8.4 MiB/s | 119.8 KiB |  00m00s
[158/215] Installing perl-File-Find-0:1 100% |   0.0   B/s |  42.5 KiB |  00m00s
[159/215] Installing perl-TermReadKey-0 100% |  64.6 MiB/s |  66.2 KiB |  00m00s
[160/215] Installing perl-lib-0:0.65-51 100% |   0.0   B/s |   8.9 KiB |  00m00s
[161/215] Installing perl-File-Copy-0:2 100% |   0.0   B/s |  20.2 KiB |  00m00s
[162/215] Installing perl-File-Which-0: 100% |   0.0   B/s |  31.4 KiB |  00m00s
[163/215] Installing perl-Error-1:0.170 100% |  78.1 MiB/s |  80.0 KiB |  00m00s
[164/215] Installing npth-0:1.8-2.fc42. 100% |  49.5 MiB/s |  50.7 KiB |  00m00s
[165/215] Installing gnupg2-0:2.4.7-2.f 100% | 245.1 MiB/s |   9.8 MiB |  00m00s
[166/215] Installing gpgme-0:1.24.2-1.f 100% |  38.7 MiB/s | 593.9 KiB |  00m00s
[167/215] Installing wget2-libs-0:2.2.0 100% | 179.1 MiB/s | 366.8 KiB |  00m00s
[168/215] Installing wget2-0:2.2.0-3.fc 100% |  61.8 MiB/s |   1.1 MiB |  00m00s
[169/215] Installing libpipeline-0:1.5. 100% |  11.9 MiB/s | 146.6 KiB |  00m00s
[170/215] Installing man-db-0:2.13.0-2. 100% |  77.0 MiB/s |   2.8 MiB |  00m00s
[171/215] Installing environment-module 100% |  62.2 MiB/s |   1.8 MiB |  00m00s
[172/215] Installing libcbor-0:0.11.0-3 100% |  77.3 MiB/s |  79.2 KiB |  00m00s
[173/215] Installing libfido2-0:1.15.0- 100% | 237.9 MiB/s | 243.6 KiB |  00m00s
[174/215] Installing emacs-filesystem-1 100% |   0.0   B/s | 544.0   B |  00m00s
[175/215] Installing munge-libs-0:0.5.1 100% |   0.0   B/s |  28.8 KiB |  00m00s
[176/215] Installing pmix-0:4.2.8-4.fc4 100% | 289.6 MiB/s |   2.0 MiB |  00m00s
[177/215] Installing prrte-libs-0:3.0.6 100% | 276.2 MiB/s |   1.7 MiB |  00m00s
[178/215] Installing prrte-0:3.0.6-6.fc 100% | 158.3 MiB/s | 162.1 KiB |  00m00s
[179/215] Installing tcsh-0:6.24.14-2.f 100% |  46.4 MiB/s |   1.3 MiB |  00m00s
[180/215] Installing orangefs-0:2.9.8-1 100% | 135.7 MiB/s |   3.1 MiB |  00m00s
[181/215] Installing openssh-0:9.9p1-12 100% |  86.1 MiB/s |   1.4 MiB |  00m00s
[182/215] Installing openssh-clients-0: 100% | 108.3 MiB/s |   2.6 MiB |  00m00s
[183/215] Installing git-core-0:2.48.1- 100% | 339.5 MiB/s |  22.7 MiB |  00m00s
[184/215] Installing git-core-doc-0:2.4 100% | 359.3 MiB/s |  17.6 MiB |  00m00s
[185/215] Installing perl-Git-0:2.48.1- 100% |  63.5 MiB/s |  65.0 KiB |  00m00s
[186/215] Installing git-0:2.48.1-3.fc4 100% |  85.4 MiB/s |  87.5 KiB |  00m00s
[187/215] Installing rocm-clang-0:18-42 100% |  77.6 MiB/s |  92.3 MiB |  00m01s
[188/215] Installing rocm-clang-devel-0 100% | 117.9 MiB/s |  21.9 MiB |  00m00s
[189/215] Installing rocm-device-libs-0 100% |  92.4 MiB/s |   3.2 MiB |  00m00s
[190/215] Installing hipcc-0:18-42.rocm 100% |  29.6 MiB/s | 605.9 KiB |  00m00s
[191/215] Installing rocm-hip-0:6.3.2-4 100% | 358.5 MiB/s |  23.3 MiB |  00m00s
[192/215] Installing rocblas-0:6.3.0-10 100% | 174.6 MiB/s |   3.8 GiB |  00m22s
[193/215] Installing rocsolver-0:6.3.0- 100% |  45.4 MiB/s | 130.2 MiB |  00m03s
[194/215] Installing hipblas-0:6.3.0-5. 100% |  94.4 MiB/s |   1.1 MiB |  00m00s
[195/215] Installing rocm-comgr-devel-0 100% |  51.0 MiB/s | 104.4 KiB |  00m00s
[196/215] Installing rocm-hip-devel-0:6 100% | 149.3 MiB/s |   2.7 MiB |  00m00s
[197/215] Installing pthreadpool-0:0.0^ 100% | 107.9 MiB/s | 110.5 KiB |  00m00s
[198/215] Installing rhash-0:1.4.5-2.fc 100% |  24.9 MiB/s | 356.4 KiB |  00m00s
[199/215] Installing libuv-1:1.50.0-1.f 100% | 278.1 MiB/s | 569.6 KiB |  00m00s
[200/215] Installing jsoncpp-0:1.9.6-1. 100% |  28.6 MiB/s | 263.1 KiB |  00m00s
[201/215] Installing cmake-data-0:4.0.0 100% | 114.8 MiB/s |   9.2 MiB |  00m00s
[202/215] Installing cmake-0:4.0.0~rc3- 100% | 149.5 MiB/s |  34.4 MiB |  00m00s
[203/215] Installing pthreadpool-devel- 100% |  97.5 MiB/s |  99.8 KiB |  00m00s
[204/215] Installing rocblas-devel-0:6. 100% | 164.3 MiB/s |   2.8 MiB |  00m00s
[205/215] Installing hipblas-devel-0:6. 100% | 163.9 MiB/s |   3.1 MiB |  00m00s
[206/215] Installing hipcc-libomp-devel 100% |   0.0   B/s | 124.0   B |  00m00s
[207/215] Installing openmpi-0:5.0.6-5. 100% | 388.1 MiB/s |   7.0 MiB |  00m00s
[208/215] Installing rocm-rpm-macros-0: 100% |   0.0   B/s |  19.4 KiB |  00m00s
[209/215] Installing wget2-wget-0:2.2.0 100% |  36.1 KiB/s | 444.0   B |  00m00s
[210/215] Installing gcc-c++-0:15.0.1-0 100% |  87.1 MiB/s |  40.8 MiB |  00m00s
[211/215] Installing libcurl-devel-0:8. 100% |  53.7 MiB/s |   1.4 MiB |  00m00s
[212/215] Installing annobin-plugin-gcc 100% |  69.4 MiB/s | 994.8 KiB |  00m00s
[213/215] Installing gcc-plugin-annobin 100% |   4.4 MiB/s |  58.8 KiB |  00m00s
[214/215] Installing langpacks-en-0:4.2 100% |   0.0   B/s | 700.0   B |  00m00s
[215/215] Installing xxd-2:9.1.1179-1.f 100% | 178.9 KiB/s |  34.4 KiB |  00m00s
Warning: skipped OpenPGP checks for 35 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.cQWPTe
Executing(%prep): /bin/sh -e /var/tmp/rpm-tmp.dnjBYu
+ 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.N2JWtX
+ 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 -DCMAKE_POLICY_VERSION_MINIMUM=3.5 -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 "2.48.1")
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 (5.3s)
-- 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 -j4 --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 -j4
/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'
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="
[  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
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'
[  1%] Building C object ggml/src/CMakeFiles/ggml-base.dir/ggml.c.o
[  3%] Building C object ggml/src/CMakeFiles/ggml-base.dir/ggml-alloc.c.o
[  2%] 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 -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
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
-- Found Git: /usr/bin/git (found version "2.48.1")
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
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
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'
[  4%] 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
[  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
gmake[2]: Leaving directory '/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/redhat-linux-build'
[  5%] Built target build_info
[  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'
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]: 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="
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'
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'
[  9%] Building C object ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o
[  8%] 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
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
[ 10%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.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/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
[ 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
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
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.
[ 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 {
      |         ^
[ 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
6 warnings generated when compiling for gfx1012.
[ 14%] 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
[ 15%] Building CXX object ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu-traits.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-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
sh: line 1: /usr/bin/rocm_agent_enumerator: No such file or directory
[ 15%] 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
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 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 {
      |         ^
6 warnings generated when compiling for gfx1010.
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 gfx1031.
[ 15%] 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/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 gfx1012.
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 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 {
      |         ^
6 warnings generated when compiling for gfx1030.
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 {
      |         ^
[ 16%] 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/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.
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 gfx1031.
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/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 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 {
      |         ^
6 warnings generated when compiling for gfx1035.
[ 17%] 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
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 {
      |         ^
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 gfx1102.
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 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/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 gfx1103.
[ 18%] 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]
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 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/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/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 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 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 {
      |         ^
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/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/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.
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 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/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/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/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 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/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/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.
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 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/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/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 {
      |         ^
/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'
[ 18%] Built target ggml-cpu
[ 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
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 gfx1152.
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/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 {
      |         ^
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 gfx1010.
6 warnings generated when compiling for gfx1200.
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/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 {
      |         ^
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 gfx1201.
6 warnings generated when compiling for gfx1012.
6 warnings generated when compiling for gfx908.
6 warnings generated when compiling for gfx900.
In file included from In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cu/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cu::41:
:
In file included from In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argmax.cuh/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/argsort.cuh::11:
:
In file included from In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh::2020:
:
/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-hip/../ggml-common.h::171171::99::  warning: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]

    171171 |  |                 ssttrruucctt  {{

            | |         ^        ^

/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:  9192: |   warning:  anonymous types declared in an anonymous union are an extension [-Wnested-anon-types] 
     str  u192c | t   { 
        |          ^ 
 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-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
:213:9: 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-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
       struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
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 gfx900.
6 warnings generated when compiling for gfx1030.
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/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: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 254{
:      |         ^
9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
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 gfx906.
6 warnings generated when compiling for gfx1031.
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/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/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/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 gfx908.
6 warnings generated when compiling for gfx1035.
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/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/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/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 gfx90a.
6 warnings generated when compiling for gfx1100.
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
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/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/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 {
      |         ^
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 gfx90a.
6 warnings generated when compiling for gfx1101.
/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 gfx942.
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/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 gfx1010.
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 gfx1102.
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 {
      |         ^
/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/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/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/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 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
6 warnings generated when compiling for gfx1103.
6 warnings generated when compiling for gfx1010.
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/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 {
      |         ^
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/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 gfx1150.
9 warnings generated when compiling for gfx1010.
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/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/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 {
      |         ^
7 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;
      |                 ^~~~~
6 warnings generated when compiling for gfx1151.
6 warnings generated when compiling for gfx1030.
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/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 {
      |         ^
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 gfx1152.
6 warnings generated when compiling for gfx1031.
/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.
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/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 |         sIn 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 |  truct {
      |         ^
       struc/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.ht {
      |         ^
: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-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an 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 {
      |         ^
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.
6 warnings generated when compiling for gfx1200.
/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;
      |                 ^~~~~
/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;
      |           ^~~~~
9 warnings generated when compiling for gfx1031.
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/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 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 {
      |         ^
6 warnings generated when compiling for gfx1100.
6 warnings generated when compiling for gfx1201.
/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 gfx1035.
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 {
      |         ^
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 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 {
      |         ^
6 warnings generated when compiling for gfx900.
/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;
      |                 ^~~~~
7 warnings generated when compiling for gfx1100.
9 warnings generated when compiling for gfx1100.
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/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 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 {
      |         ^
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/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
/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 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/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 gfx1101.
6 warnings generated when compiling for gfx1103.
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 {
      |         ^
/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/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 {
      |         ^
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/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 gfx1150.
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 {
      |         ^
7 warnings generated when compiling for gfx1102.
/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
      |         ~~~~~~~~~~ ^ ~~~~
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]
/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;
      |                 ^~~~~
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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 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 {
      |         ^
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 {
      |         ^
6 warnings generated when compiling for gfx90a.
/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/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 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 gfx1152.
7 warnings generated when compiling for gfx1103.
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/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 {
      |         ^
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/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/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 {
      |         ^
9 warnings generated when compiling for gfx1151.
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;
      |           ^~~~~
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
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.hIn file included from :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: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 {
      |         ^
7 warnings generated when compiling for gfx1150.
/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 {
      |         ^
6 warnings generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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 gfx1152.
/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/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 {
      |         ^
17 warnings generated when compiling for gfx1010.
/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 {
      |         ^
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 gfx900.
/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 {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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 gfx1200.
7 warnings generated when compiling for gfx1151.
/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.
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 {
      |         ^
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.
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/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;
      |                 ^~~~~
/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;
      |           ^~~~~
9 warnings generated when compiling for gfx1201.
/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 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 {
      |         ^
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/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 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/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;
      |                 ^~~~~
/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);
      |                   ^~~~~~~~~~
9 warnings generated when compiling for gfx900.
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/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;
      |           ^~~~~
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 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/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;
      |                 ^~~~~
7 warnings generated when compiling for gfx1200.
/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);
      |                   ^~~~~~~~~~
9 warnings generated when compiling for gfx906.
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/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 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/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/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
/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;
      |                 ^~~~~
/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.
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/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 gfx1201.
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/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);
      |                   ^~~~~~~~~~
/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;
      |                 ^~~~~
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/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 {
      |         ^
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/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 host.
[ 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
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);
      |                   ^~~~~~~~~~
7 warnings generated when compiling for gfx900.
17 warnings generated when compiling for gfx1102.
/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/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 {
      |         ^
9 warnings generated when compiling for gfx90a.
/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/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/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/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/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 gfx1103.
13 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 gfx942.
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/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 {
      |         ^
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);
      |                   ^~~~~~~~~~
/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 |     17 warnings generated when compiling for gfx1150.
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/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/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 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
/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;
      |           ^~~~~
13 warnings generated when compiling for gfx1012.
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);
      |                   ^~~~~~~~~~
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/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 {
      |         ^
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/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 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: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;
      |                                 ^
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-cuda/convert.cu/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
588:5:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
 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/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
/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/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 {
      |         ^
17 warnings generated when compiling for gfx1152.
13 warnings generated when compiling for gfx1030.
/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.
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/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);
      |                   ^~~~~~~~~~
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;
      |                                 ^
17 warnings generated when compiling for gfx1200.
/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/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/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 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;
      |           ^~~~~
13 warnings generated when compiling for gfx1031.
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/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 {
      |         ^
7 warnings generated when compiling for gfx90a.
17 warnings generated when compiling for gfx1201.
/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 gfx1031.
/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/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/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/binbcast.cu:359:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  359 |         } break;
      |           ^~~~~
/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/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 {
      |         ^
17 warnings generated when compiling for gfx900.
13 warnings generated when compiling for gfx1035.
/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.
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/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 kerIn 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 {
      |         ^
ne/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
l_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 {
      |         ^
= ggml_nelem/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
en/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
ts(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.
/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/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/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
7 warnings generated when compiling for gfx1100.
/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/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:
13In file included from  warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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,
      |                                                                            ^
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/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);
      |                   ^~~~~~~~~~
7 warnings generated when compiling for host.
17 warnings generated when compiling for gfx908.
[ 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
/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 {
      |         ^
/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>;
      |                    ^
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: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/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/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/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.
13 warnings generated when compiling for gfx1101.
/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.
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/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 {
      |         ^
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/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.
/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/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 {
      |         ^
/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/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: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>;
      |                    ^
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/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 gfx1102.
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 {
      |         ^
6 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 gfx942.
/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 {
      |         ^
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 {
      |         ^
/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>;
      |                    ^
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/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.
[ 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
/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 gfx1103.
7 warnings generated when compiling for gfx1151.
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/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 {
      |         ^
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 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: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/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: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>;
      |                    ^
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/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 {
      |         ^
13 warnings generated when compiling for gfx1150.
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/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;
      |             ^~~~~
6 warnings generated when compiling for gfx1012.
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 {
      |         ^
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 {
      |         ^
/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/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/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;
      |             ^~~~~
6 warnings generated when compiling for gfx1030.
6 warnings generated when compiling for gfx1035.
7 warnings generated when compiling for gfx1201.
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/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 {
      |         ^
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;
      |                                 ^
6 warnings generated when compiling for gfx1031.
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>;
      |                    ^
6 warnings generated when compiling for gfx1100.
13 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/cross-entropy-loss.cuIn 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 {
      |         ^
:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         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-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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.
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/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 {
      |         ^
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 {
      |         ^
/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;
      |             ^~~~~
6 warnings generated when compiling for gfx1101.
7 warnings generated when compiling for gfx908.
6 warnings generated when compiling for gfx1100.
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/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/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/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/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 gfx90a.
/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>;
      |                    ^
6 warnings generated when compiling for gfx1101.
/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>;
      |                    ^
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/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 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 {
      |         ^
/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/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 {
      |         ^
7 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/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>;
      |                    ^
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/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 {
      |         ^
/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 gfx942.
13 warnings generated when compiling for gfx900.
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/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 {
      |         ^
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 {
      |         ^
/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 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
/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);
      |         ^
6 warnings generated when compiling for gfx1150.
/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>;
      |                    ^
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 {
      |         ^
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/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 gfx1151.
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/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 {
      |         ^
6/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
 warnings generated when compiling for gfx1151.
/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/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 gfx1012.
13 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 {
      |         ^
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/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/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 {
      |         ^
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 {
      |         ^
6 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>;
      |                    ^
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/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 {
      |         ^
13 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/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/convert.cu6 warning: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:s9:  generated when compiling for gfx1201.
warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h: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/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 {
      |         ^
/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>;
      |                    ^
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/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 {
      |         ^
13 warnings generated when compiling for gfx90a.
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 {
      |         ^
6 warnings generated when compiling for gfx1100.
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/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/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 {
      |         ^
/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>;
      |                    ^
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/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 gfx1101.
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/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.
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:In file included from 254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/diagmask.cu:1:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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/diagmask.cuh:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h: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.
/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);
      |         ^
6 warnings generated when compiling for gfx1102.
/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/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 {
      |         ^
13 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/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/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/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 gfx906.
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/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 {
      |         ^
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 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/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/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 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 {
      |         ^
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 {
      |         ^
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 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
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/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/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/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-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) {
      |                         ~ ^ ~~~~~~~~~~~
30 warnings generated when compiling for gfx1010.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((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);
      |             ^
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-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);
      |         ^
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/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/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/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);
      |             ^
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/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 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/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 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
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.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 {
      |         ^
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.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);
      |             ^
6 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-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/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-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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.hIn file included from :213:9: 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/fattn-tile-f32.cu struct {
      :|         ^
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);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an 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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((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 gfx1012.
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);
      |         ^
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/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/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 {
      |         ^
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/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);
      |             ^
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/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.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/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 {
      |         ^
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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h n:298e:190:, 
warning:       | anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]                  ^

/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-vec-f16.cuh:28  :29819 | :   warning:  unused parameter 'ne13' [-Wunused-parameter] 
    st   r28u | c t   { 
        |          ^ 
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);
      |             ^
6 warnings generated when compiling for gfx942.
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/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.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.
[ 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
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 gfx1031.
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/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 gfx1010.
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: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/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/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 gfx1035.
30 warnings generated when compiling for gfx1030.
/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 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-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.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-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-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 ) &= ftz_mask;
      |                        ^
/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:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh 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) {
      |                   ^
:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_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: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_/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]
_    24K | , 
        |                                    ^ 
   const int ne00,
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:16/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu::3527:: 19warning: :unused parameter 'V' [-Wunused-parameter] 
warning: unused parameter 'ne03' [-Wunused-parameter]
   27 |     16  |             c o ncsotn sitn tc hnaer0 3*, 
_      _| r                  ^e
str/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cui:c28t:_19_:  Vwarning: ,unused parameter 'ne10' [-Wunused-parameter]

      |                                   ^
   28 |  /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh : 17 : 35 :   warning: cunused parameter 'mask' [-Wunused-parameter]o
nst in   t17  | n e 1 0 , 
        |                    ^c
ons/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cut: 31c:h19a:r  warning: *unused parameter 'ne13' [-Wunused-parameter] 
__res   t31r | i c t _ _   m a scko,n
s      t|                                    ^i
nt n/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuhe:1183:,35
:       | warning:                   ^unused parameter 'dst' [-Wunused-parameter]

/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:32:   1918: |   warning:  unused parameter 'ne31' [-Wunused-parameter] 
        32f | l o a t          c o*n s_t_ rienstt rniec3t1_,_
       d| s                  ^t
,
      /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu| :                                  ^33
:19: /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuhwarning: :unused parameter 'nb31' [-Wunused-parameter]19
:35: warning:    unused parameter 'dst_meta' [-Wunused-parameter]33
 |           19  | c o n s t   i n tf lnoba3t12, 
        |                    ^ 
* _/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu_:r36e:s19t:r iwarning: cunused parameter 'nb03' [-Wunused-parameter]t
__ d   s36t | _ m e t a , 
        | c                                  ^o
nst/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh :i20n:t21 :n bwarning: 0unused parameter 'scale' [-Wunused-parameter]3
,
      |                   ^
   20 |  /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu : 39 : 19 :   warning: cunused parameter 'nb13' [-Wunused-parameter]o
nst    f39l | o a t   s c a l ec,o
n      s| t                    ^ 
int/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh :n21b:1213:, 
warning:       unused parameter 'max_bias' [-Wunused-parameter]| 
                  ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu:   4021: | 19 :   warning:  unused parameter 'nb21' [-Wunused-parameter] 
   co   n40s | t   f l o a t   mcaoxn_sbti aisn,t
       n| b                    ^2
1,
      /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh| :                  ^22
:21: /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cuwarning: :unused parameter 'm0' [-Wunused-parameter]41
:19: warning:    unused parameter 'nb22' [-Wunused-parameter]22
 |        41  |       c o n s t  cfolnosatt  imn0t, 
n      b| 2                    ^2
,
      /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh| :                  ^23
:21:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu :warning: 42unused parameter 'm1' [-Wunused-parameter]:
19: warning: unused parameter 'nb23' [-Wunused-parameter]   
23 |        42  |       c o n s t  cfolnosatt  imn1t, 
n      b| 2                    ^3
,
      | /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh                  ^:
24:24:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu :warning: 43unused parameter 'n_head_log2' [-Wunused-parameter]:
19: warning: unused parameter 'ne0' [-Wunused-parameter]
   24 |      43  |           c o n scto nusitn ti3n2t_ tn en0_,h
e      a| d                  ^_
log/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu2:,44
:      19| :                       ^ 
warning: unused parameter 'ne1' [-Wunused-parameter]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:25:   2144: |   warning:  unused parameter 'logit_softcap' [-Wunused-parameter] 
     c   o25n | s t   i n t   n ec1o,n
s      t|                    ^f
loa/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cut: 45l:o19g:i twarning: _unused parameter 'ne2' [-Wunused-parameter]s
oftc   a45p | , 
        |                     ^ 
   /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh :c26o:n19s:t  warning: iunused parameter 'ne00' [-Wunused-parameter]n
t ne   226, | 
        |                    ^ 
   /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-tile-f16.cu :c46o:n19s:t  warning: iunused parameter 'ne3' [-Wunused-parameter]n
t ne0   046, | 
        |                    ^ 
   /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-wmma-f16.cuh :c27o:n19s:t  warning: iunused parameter 'ne01' [-Wunused-parameter]n
t ne   327) |   { 
        |                    ^ 
  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-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.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-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/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-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);
      |         ^
/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 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/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;
      |             ^~~~~
7 warnings generated when compiling for gfx1031.
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/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 {
      |         ^
30 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 {
      |         ^
/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;
      |                     ^~~~~
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 gfx1035.
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 {
      |         ^
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-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/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 {
      |         ^
30 warnings generated when compiling for gfx1031.
/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) {
      |                                               ^
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/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/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 {
      |         ^
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/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-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);
      |             ^
/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-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.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 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:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-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 gfx1150.
/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 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/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);
      |             ^
30 warnings generated when compiling for gfx1031.
/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 gfx1151.
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-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/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-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.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<128, 1>' 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) {
      |                                               ^
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 gfx1152.
7 warnings generated when compiling for gfx1150.
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.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-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 {
      |         ^
/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;
      |                     ^~~~~
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-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) {
      |                   ^
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-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);
      |         ^
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;
      |             ^~~~~
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 gfx1152.
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 gfx1201.
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 {
      |         ^
30 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 gfx1200.
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-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: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 {
      |         ^
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) {
      |                   ^
80 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: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);
      |         ^
/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/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 {
      |         ^
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.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 gfx906.
/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-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 {
      |         ^
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/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.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/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;
      |             ^~~~~
7 warnings generated when compiling for gfx906.
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 {
      |         ^
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) {
      |                         ~ ^ ~~~~~~~~~~~
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) {
      |                                               ^
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 gfx908.
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/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/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-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) {
      |                   ^
/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 {
      |         ^
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);
      |         ^
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/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 {
      |         ^
30 warnings generated when compiling for gfx1101.
/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 {
      |         ^
7 warning/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
s 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.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/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/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-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);
      |         ^
80 warnings generated when compiling for gfx942.
/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 gfx942.
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: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 |      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) {
      |                   ^
   struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an 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: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 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
/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 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
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/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/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/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/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) {
      |                   ^
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/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/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 gfx1010.
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);
      |         ^
30 warnings generated when compiling for gfx1102.
/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/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/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) {
      |                         ~ ^ ~~~~~~~~~~~
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/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((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/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;
      |             ^~~~~
6 warnings generated when compiling for gfx1012.
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/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 {
      |         ^
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/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 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-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) {
      |                   ^
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/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/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);
      |         ^
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 {
      |         ^
/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 gfx1035.
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/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 gfx1031.
/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/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 {
      |         ^
29 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: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/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/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/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;
      |             ^~~~~
30 warnings generated when compiling for gfx1103.
29 warnings generated when compiling for gfx1101.
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-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/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) {
      |                                                                                                   ^
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-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-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((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-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-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) {
      |                   ^
/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;
      |               ^~~
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);
      |             ^
/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);
      |         ^
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);
      |         ^
/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.
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: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);
      |         ^
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 {
      |         ^
/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;
      |             ^~~~~
30 warnings generated when compiling for gfx1150.
29 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 {
      |         ^
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:
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-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) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((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/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-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) {
      |                   ^
/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;
      |               ^~~
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);
      |             ^
/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 |  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);
      |             ^
   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/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/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 gfx1150.
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/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;
      |             ^~~~~
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/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 {
      |         ^
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/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/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 gfx1152.
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 {
      |         ^
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/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);
      |         ^
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 {
      |         ^
/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.
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 |         *((uint32In file included from _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);
      |         ^
/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 gfx1150.
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;
      |             ^~~~~
30 warnings generated when compiling for gfx1151.
29 warnings generated when compiling for gfx1201.
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/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/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/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/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;
      |             ^~~~~
6 warnings generated when compiling for gfx1151.
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/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 gfx906.
30 warnings generated when compiling for gfx1152.
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/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;
      |               ^~~
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/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 {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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/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/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) {
      |                   ^
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/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 gfx1152.
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/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;
      |               ^~~
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 {
      |         ^
/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);
      |         ^
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 {
      |         ^
/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.
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);
      |         ^
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/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 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:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const 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;
      |             ^~~~~
30 warnings generated when compiling for gfx1200.
6 warnings generated when compiling for gfx900.
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/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/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/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/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) {
      |                                                                                                         ^
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/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;
      |               ^~~
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);
      |         ^
/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/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) {
      |                   ^
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-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_logit29_softcap>(ctx, dst);
      |         ^
 warnings generated when compiling for host.
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, ds[ 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
t);
      |         ^
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 gfx1010.
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 gfx908.
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/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 {
      |         ^
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/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 gfx1030.
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 gfx1201.
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) {
      |                   ^
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/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/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/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 {
      |         ^
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/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/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 gfx1035.
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 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 gfx942.
6 warnings generated when compiling for gfx1101.
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/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 {
      |         ^
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/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
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 gfx1102.
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-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/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/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/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 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/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 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/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 gfx1151.
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/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:
/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.
30 warnings generated when compiling for gfx906.
6 warnings generated when compiling for gfx1152.
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/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/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/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;
      |             ^~~~~
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) {
      |                                               ^
15 warnings generated when compiling for gfx1035.
/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: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) {
      |                         ~ ^ ~~~~~~~~~~~
6 warnings generated when compiling for gfx1200.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((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/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) {
      |                                                                                                   ^
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-cuda/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |          ^
1053 |     const int * __restrict__ 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-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, 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 gfx1201.
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) {
      |                                                                                                   ^
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 {
      |         ^
/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.
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/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;
      |             ^~~~~
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 {
      |         ^
15 warnings generated when compiling for gfx1102.
6 warnings generated when compiling for gfx906.
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/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;
      |             ^~~~~
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 {
      |         ^
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/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 {
      |         ^
30 warnings generated when compiling for gfx908.
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-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/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/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/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_I16K4In 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 {
      |         ^
 & mma_A, const mma_int_B_J8K4 /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
& mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  /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) {
      |                                                                                                         ^
213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an 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:
/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;
      |             ^~~~~
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 {
      |         ^
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/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 gfx90a.
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/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;
      |             ^~~~~
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 {
      |         ^
15 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/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;
      |             ^~~~~
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 {
      |         ^
15 warnings generated when compiling for gfx1152.
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/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;
      |             ^~~~~
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 {
      |         ^
15 warnings generated when compiling for gfx1200.
30 warnings generated when compiling for gfx90a.
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/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/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;
      |             ^~~~~
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) {
      |                         ~ ^ ~~~~~~~~~~~
15 warnings generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((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/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/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/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/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/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]
6 warnings generated when compiling for gfx1010.
 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.
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/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 gfx906.
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/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 {
      |         ^
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/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 gfx1030.
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/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/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.
30 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/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/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/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) {
      |                                                                                                   ^
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/mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmq.cuh/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) {
      |                                               ^
:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/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) {
      |                   ^
/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;
      |             ^~~~~
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);
      |             ^
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/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/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 {
      |         ^
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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:9:: 213note: :in instantiation of function template specialization 'launch_fattn_tile_f16_64_128<32, 1, false>' requested here
9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  348 |     213  |           l a un csht_rfuacttt n{_
t      i| l        ^e
_f16_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-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an 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 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
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/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 {
      |         ^
30 warnings generated when compiling for gfx942.
/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/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 {
      |         ^
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/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 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/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 {
      |         ^
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/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/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 gfx1101.
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/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/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/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 {
      |         ^
30 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
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 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 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/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/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/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.
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 gfx1030.
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/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/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.
6 warnings generated when compiling for gfx1031.
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 {
      |         ^
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 {
      |         ^
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 gfx1100.
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 {
      |         ^
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.
6 warnings generated when compiling for gfx1100.
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/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/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 {
      |         ^
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 gfx1101.
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 gfx1152.
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 {
      |         ^
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 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 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 gfx1200.
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.
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/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/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/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.
6 warnings generated when compiling for gfx1151.
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 {
      |         ^
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/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 gfx1152.
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/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 {
      |         ^
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.
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/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 {
      |         ^
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/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/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.
6 warnings generated when compiling for gfx906.
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 {
      |         ^
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/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.
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/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.
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 {
      |         ^
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 {
      |         ^
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.
6 warnings generated when compiling for gfx908.
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/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/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/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 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/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/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.
6 warnings generated when compiling for host.
[ 37%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.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/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
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/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/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.
6 warnings generated when compiling for gfx942.
6 warnings generated when compiling for gfx942.
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/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/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/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 gfx1012.
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.
[ 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
6 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
/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);
      |     ^
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-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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<ty/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.hpe:, 2542:>9<:< <warning: banonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
lock_n  u254m | s ,  b l o c k _ dsitmrsu,c t0 ,{ 
s      t| r        ^e
am>>>(vx, vy, dst, 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-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
  298 |         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 (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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/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/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 {
      |         ^
6 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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.
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 {
      |         ^
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 {
      |         ^
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 {
      |         ^
/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.
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 {
      |         ^
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 {
      |         ^
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/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/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.
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 {
      |         ^
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: 6 warnings generated when compiling for gfx1100.
anonymous types declared in an anonymous union are an 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/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.
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 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 {
      |         ^
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/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.
6In file included from  warnings generated when compiling for gfx1035.
/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 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 {
      |         ^
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/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/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.
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/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/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/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/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 gfx1101.
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/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) {
      |                        ~~~~~~~~~~ ^ ~~~~
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 {
      |         ^
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/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 gfx1102.
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/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) {
      |                        ~~~~~~~~~~ ^ ~~~~
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 {
      |         ^
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/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.
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/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 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 {
      |         ^
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/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 {
      |         ^
6 warnings generated when compiling for gfx1200.
/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 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/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/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 {
      |         ^
6 warnings generated when compiling for gfx1201.
/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/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/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/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 {
      |         ^
6 warnings generated when compiling for gfx900.
/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/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/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/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 {
      |         ^
6 warnings generated when compiling for gfx1200.
6 warnings generated when compiling for gfx906.
/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/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/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/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 {
      |         ^
6 warnings generated when compiling for gfx1201.
6 warnings generated when compiling for gfx908.
/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/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/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/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 {
      |         ^
6 warnings generated when compiling for gfx900.
/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) {
      |                        ~~~~~~~~~~ ^ ~~~~
6 warnings generated when compiling for gfx90a.
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/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/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 {
      |         ^
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:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |        /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h :s192t:r9u:c twarning:  anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]{

      |         ^
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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) {
      |                        ~~~~~~~~~~ ^ ~~~~
6 warnings generated when compiling for gfx906.
8 warnings generated when compiling for gfx908.
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/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/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/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/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.
6 warnings generated when compiling for gfx908.
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/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/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 {
      |         ^
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:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h |         struct {
      |         ^
: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-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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 gfx90a.
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
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/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 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]
/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 {
      |         ^
  213 |   /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
      struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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/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;
      |             ^~~~~
6 warnings generated when compiling for gfx90a.
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/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) {
      |                        ~~~~~~~~~~ ^ ~~~~
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 {
      |         ^
8 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/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 {
      |         ^
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/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/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 {
      |         ^
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/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 {
      |         ^
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/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/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/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 {
      |         ^
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/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/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 {
      |         ^
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/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/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.
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/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;
      |             ^~~~~
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 {
      |         ^
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/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/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/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/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.
6 warnings generated when compiling for gfx1035.
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/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/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/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;
      |             ^~~~~
6 warnings generated when compiling for gfx1100.
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/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 {
      |         ^
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/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) {
      |                                                                                                         ^
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/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/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/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 {
      |         ^
6 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/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/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 {
      |         ^
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/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/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/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 {
      |         ^
6 warnings generated when compiling for gfx1035.
293 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: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;
      |             ^~~~~
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 {
      |         ^
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/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/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;
      |             ^~~~~
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 {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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_cIn 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 {
      |         ^
uda_block] = {0.0/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
f};
      |                                                ^~~~
      |                                                {   }
/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/mmvq.cu:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 1>' requested here
  176 |           /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
  mul_mat_vec_q<type, 1><<<block/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
_nums, block/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
_dims, 0, stream>>>(vx, vy, dst, 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/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.
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/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/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/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;
      |             ^~~~~
6 warnings generated when compiling for gfx1200.
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/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 {
      |         ^
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/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 {
      |         ^
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/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/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/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 {
      |         ^
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/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;
      |             ^~~~~
6 warnings generated when compiling for gfx1101.
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/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/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/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/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/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;
      |             ^~~~~
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.
15 warnings generated when compiling for gfx90a.
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 {
      |         ^
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/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/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.
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/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/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 {
      |         ^
161 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: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;
      |             ^~~~~
6 warnings generated when compiling for gfx942.
6 warnings generated when compiling for gfx1103.
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/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)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
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 {
      |         ^
/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 hereIn 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 {
      |         ^

  230 |     mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(vx, vy/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types],
  192 |         struct {
      |         ^
 dst, ncols_x, nrows_x, nrow/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
s_y, ncols_y, nrows_dst, stream);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in 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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
: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:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         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_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-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^

  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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/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;
      |             ^~~~~
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
15 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/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/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.
6 warnings generated when compiling for gfx1150.
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/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 {
      |         ^
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/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.
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 {
      |         ^
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.
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 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 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 {
      |         ^
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 gfx1100.
6 warnings generated when compiling for gfx1031.
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 {
      |         ^
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/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/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.
161 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/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 {
      |         ^
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.
/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);
      |     ^
6 warnings generated when compiling for gfx1103.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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/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 {
      |         ^
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 {
      |         ^
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 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 gfx1101.
6 warnings generated when compiling for gfx1151.
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 {
      |         ^
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 {
      |         ^
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.
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 {
      |         ^
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/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.
6 warnings generated when compiling for gfx900.
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 {
      |         ^
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 {
      |         ^
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.
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 {
      |         ^
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.
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 gfx1151.
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 {
      |         ^
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 {
      |         ^
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/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/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:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_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_TYIn 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 {
      |         ^
PE_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-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in 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: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-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
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-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_b/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
lock] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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><<</builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.hbl:o213c:k9_:n ums, warning: banonymous types declared in an anonymous union are an extension [-Wnested-anon-types]l
ock_dims  ,213  | 0,   s t r e a m >s>t>r(uvcxt,  {v
y      ,|          ^d
st, 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-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         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:182:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ4_NL, 3>' requested here
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h  :281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
182 |             mul/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, 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 gfx90a.
6 warnings generated when compiling for gfx908.
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 {
      |         ^
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/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/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
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 {
      |         ^
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.
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 {
      |         ^
6 warnings generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
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/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/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.
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 {
      |         ^
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.
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 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/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/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.
161 warnings generated when compiling for gfx1100.
6 warnings generated when compiling for gfx1100.
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 {
      |         ^
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 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:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |      /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
   struct {
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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);
      |     ^
6 warnings generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
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: /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         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: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-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in 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: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, str/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.he:a254m:)9;:
       warning: |     ^anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]

  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in 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:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9:    80warning:  | anonymous types declared in an anonymous union are an extension [-Wnested-anon-types] 
   float   t298m | p [ n c o l s _ ys]t[rruocwts _{p
e      r| _        ^c
uda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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 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 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
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 {
      |         ^
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 {
      |         ^
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.
6 warnings generated when compiling for gfx1103.
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 {
      |         ^
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 {
      |         ^
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.
6 warnings generated when compiling for gfx1150.
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/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 {
      |         ^
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 {
      |         ^
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.
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/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 {
      |         ^
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 {
      |         ^
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.
6 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/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 {
      |         ^
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 {
      |         ^
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.
6 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/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 {
      |         ^
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 {
      |         ^
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.
6 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/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 {
      |         ^
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 {
      |         ^
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.
6 warnings generated when compiling for gfx900.
161 warnings generated when compiling for gfx1101.
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 {
      |         ^
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 {
      |         ^
In file included from 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 |      /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/unary.cu   struct {
      |         ^
: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:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h: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.
/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[nc6ols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
 warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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/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 {
      |         ^
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.
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 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 {
      |         ^
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.
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 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 {
      |         ^
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.
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 {
      |         ^
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/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.
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 {
      |         ^
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/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.
[ 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
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 {
      |         ^
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/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.
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/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/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 {
      |         ^
/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/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.
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/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/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 {
      |         ^
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/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/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 gfx906.
/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.
161 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 {
      |         ^
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/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 gfx908.
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/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.
/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_matIn file included from _vec_q<type, 8><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, 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/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 |         st/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.curuct {
      |         ^
:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in 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/mmvq.cu:176:13: note: in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_Q3_K, 1>' requested here
  176 |             /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
mul_mat_vec_q<type, 1><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, 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-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in 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:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
: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        ^
:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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.cuIn file included from :80:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/tsembd.cu48::1 :
warning: In file included from suggest braces around initialization of subobject [-Wmissing-braces]/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]   
80 |     float tmp  [171n | c o l s _ y ] [ rsotwrsu_cpte r{_
c      u| d        ^a
_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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-hip/../ggml-common.h:192:9: warning: anonymous types declared in 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/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h      | :                                               {   }213
:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu213: | 182 : 13 :   note:  in instantiation of function template specialization 'mul_mat_vec_q<GGML_TYPE_IQ3_S, 3>' requested here 
 struct {
      |         ^
  182 |             mul_mat_vec_q<type, 3><<<block_nums, block_dims, 0, stream>>>(vx, vy, dst, 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-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in 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:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_blo/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
ck] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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/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 gfx900.
6 warnings generated when compiling for gfx90a.
/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/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 {
      |         ^
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 {
      |         ^
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 gfx906.
6 warnings generated when compiling for gfx90a.
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/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 {
      |         ^
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 {
      |         ^
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 gfx942.
7 warnings generated when compiling for gfx1102.
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/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 {
      |         ^
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: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-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
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 {
      |         ^
/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 gfx1103.
6 warnings generated when compiling for host.
[ 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
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/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 {
      |         ^
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 {
      |         ^
/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 gfx1150.
6 warnings generated when compiling for gfx1010.
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/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.
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.hIn 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 {
      |         ^
:298/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
:9/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h: :warning: 281:anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]9
: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
    281298 |  |                 ssttrruucct {
      |         ^
t {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h: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 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 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 {
      |         ^
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 {
      |         ^
6 warnings generated when compiling for host.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
[ 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 gfx1030.
/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.
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
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/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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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);
      |     ^
161 warnings generated when compiling for gfx1103.
/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 gfx1201.
6 warnings generated when compiling for gfx1031.
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/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 {
      |         ^
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/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]/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/upscale.cu[rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
: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);
      |                                     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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 here7 warnings generated when compiling for gfx900.

  244 | In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
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:    mul_mat_vec_q_cuda<GGML_TYPE_Q2_K>(vx, vy, dst, ncols_x, nrows_x, nrows_y, ncols_y, nrows_dst, stream);
      |     ^
192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         s/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cutruct {
      |         ^
:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         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: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-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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 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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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/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 {
      |         ^
64 warnings generated when compiling for gfx1012.
7 warnings generated when compiling for gfx906.
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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
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/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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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) {
      |                   ^
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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 {
      |         ^
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 {
      |         ^
64 warnings generated when compiling for gfx1030.
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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
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/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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 gfx1102.
/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.
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/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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
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/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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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/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 {
      |         ^
64 warnings generated when compiling for gfx1035.
/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.
[ 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
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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
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 {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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 gfx1100.
6 warnings generated when compiling for gfx1151.
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/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 {
      |         ^
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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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 {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         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_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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 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/mmvq.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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_blocIn file included from k,/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cuh :t1r:
uIn file included from e/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/common.cuh,: 20t:
rue/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h);:
171      :| 9        ^:
 warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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<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 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:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_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_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 {
      |         ^
dims, 0/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
, stream/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
>>>(vx, vy, dst, 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-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
61 warnings generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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 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-cpb32.cu:3:
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/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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
6 warnings generated when compiling for gfx1201.
64 warnings generated when compiling for gfx1102.
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/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-cpb32.cu:3:
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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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, /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuhparallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |         ^
:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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:
/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]
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
   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<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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 gfx1103.
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/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-cpb32.cu:3:
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 {
      |         ^
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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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;
In file included from       |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh: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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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, pa/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuhrallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_caseIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
(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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5:In file included from  note: /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cuin instantiation of function template specialization 'flash_attn_combine_results<96, 1>' requested here:
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]
  704 |     flash   _14a | tt n _ c o m b i nceo_nrsets uclhtasr< D*,  _p_arreasltlreilc_tb_l_o cQk,s
>      
|                                         ^| 
    ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:15:35:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh :warning: 505unused parameter 'K' [-Wunused-parameter]:
5: note: in instantiation of function template specialization 'launch_fattn<96, 1>' requested here
   15 |         co  n505s | t   c h alra u*n c_h__rfeastttrni<cDt,_ _p aKr,a
l      l| e                                  ^l
_blo/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuhc:k16s:>35(:c twarning: xunused parameter 'V' [-Wunused-parameter],
 dst,    f16a | t t n _ k e r n eclo,n sntw acrhpasr,  *c o_l_sr_epsetrr_ibclto_c_k ,V ,t
r      u| e                                  ^,
 tru/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuhe:)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<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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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/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 gfx1150.
61 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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
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/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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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]
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_kerIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
nel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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;
      |                        ^In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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

/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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) &= ftIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
z_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 |      In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
   *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |     ^
6 warnings generated when compiling for gfx90a.
61 warnings generated when compiling for gfx1100.
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/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-cpb32.cu:3:
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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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<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) &/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
=/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
 ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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:In file included from /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) {
      |                   ^
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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<128, 2>' requested here
  704 |     flash_attn_combine_results<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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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<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.cuhIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<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.
61 warnings generated when compiling for gfx1101.
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/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-cpb32.cu:3:
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 {
      |         ^
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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
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 {
      |         ^
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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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>
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/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-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) {
      |                   ^
554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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/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/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 {
      |         ^
61 warnings generated when compiling for gfx1102.
/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;
      |             ^~~~~
64 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: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);
      |     ^
6 warnings generated when compiling for host.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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 |    [ 50%] Building CXX object ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu.o
 mul_mat_vec_q_cuda<GGML_TYPE_Q5_1>(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/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
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
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-kqhalf-cpb16.cu:3:
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: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |        In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
 struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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<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_blockIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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) {
      |                         ~ ^ ~~~~~~~~~~~
, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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:
/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<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-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
61 warnings generated when compiling for gfx1103.
64 warnings generated when compiling for gfx1201.
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-cpb16.cu:3:
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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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-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-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
61 warnings generated when compiling for gfx1150.
64 warnings generated when compiling for gfx900.
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-cpb16.cu:3:
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, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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:
/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<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-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
61 warnings generated when compiling for gfx1151.
64 warnings generated when compiling for gfx906.
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-cpb16.cu:3:
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 {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
*((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
, 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-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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;
      |                        ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2/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-common.cuh704::5554:: 24note: in instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested here
: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  704 |     fl  a554s | h_ a t t n _ c o m*b(i(nuei_nrte3s2u_ltt s*<)D ,& KpQa_rmaalxl_eslc_ablleo)c k&s=> 
f      t| z    ^_
mask;
      |                        ^/builddir/build/BUILD/llama-cpp-b4580-build/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
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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
  491 |        704  |     l a ufnlcahs_hf_aattttnn<_Dc,o mpbairnael_lreels_ublltosc<kDs,> (pcatrxa,l ldeslt_,b lfoactktsn>_
k      e| r    ^n
el, nw/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuha:r491p:s9,:  cnote: oin instantiation of function template specialization 'launch_fattn<64, 2>' requested herel
s_per_block, t  r491u | e ,   t r u e ) ;l
a      u| n        ^c
h_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<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;
      | In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
                       ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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 In file included from /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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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<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<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-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<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<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:505In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3::
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
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<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<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<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<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<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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |     ^
61 warnings generated when compiling for gfx1152.
64 warnings generated when compiling for gfx908.
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-cpb16.cu:3:
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 {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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]
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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 |     fo   15r (int l |  =         0; consl <t char * si __restzeof(intr); ++lic) {
t__ 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]
         |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l =17 1; l |      < sizeof(int); ++l) {
    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:       |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |      unused parameter 'dst' [-Wunused-parameter]   for (int
    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]
l = 0; l <    sizeof(int); +19 |   +l) {
      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-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-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 hereIn file included from 
  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-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh: *((uint32_554t *) &KQ_max_sca:24le) &=: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
 ftz_mask;
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
      /builddir/build/BUILD/llama-cpp-b4580-build/llama.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-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, In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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<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<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<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, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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<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<112, 4>' requested here
  704In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |     ^
_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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.cuhIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
: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);
      |         ^
61 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: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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/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-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
554:24  :491  | warning:   cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual] 
     launch_fatt  n554< | D,   p a r a l l e*l(_(bulionctk3s2>(_ctt x*,)  d&sKtQ,_ mfaaxt_tsnc_akleer)n e&l=,  fntwza_rmpass,k ;c
o      l| s                       ^_
per_block, tru/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuhe:,704 :t5r:u enote: )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]
In file included from   /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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 {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
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:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         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-hip/../ggml-common.h::254192::99::  warning: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]

    254192 |  |                 ssttrruucctt  {{

            | |         ^        ^

/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu::33:
:
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.cuh/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh::22:
:
/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-common.cuh::318318::2323::  warning: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]

    318318 |  |         ffoorr  ((iinntt  ll  ==  00;;  ll  <<  ssiizzeeooff((iinntt));;  ++++ll))  {{

            | |                     ~ ^ ~~~~~~~~~~~                    ~ ^ ~~~~~~~~~~~

/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-common.cuh::325325::2323::  warning: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]

    325325 |  |         ffoorr  ((iinntt  ll  ==  11;;  l l << s siizezoefo(if(nit)n;t) ;++ +l+) l){ 
{      | 
                          ~ ^ ~~~~~~~~~~~| 
                    ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh27::341 :warning: 27comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]:
 warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |    341  |             f o rf o(ri n(ti nlt  =l  0=;  0l;  <l  s<i zseiozfe(oifn(ti)n;t )+;+ l+)+ l{)
       {| 
                        ~ ^ ~~~~~~~~~~~      
|                         ~ ^ ~~~~~~~~~~~
/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-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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-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) {
      |                   ^
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<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-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<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<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<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<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<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<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<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-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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<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);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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 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 {
      |         ^
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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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-kqfloat-cpb16.cu:3:
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 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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_In file included from max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |     /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
    *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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:
/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]
   In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
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<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);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 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-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-kqfloat-cpb16.cu:3:
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 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:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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 |         *In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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) {
      |                         ~ ^ ~~~~~~~~~~~
((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^

  704 |     flash_attn_comb/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
ine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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:
/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<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);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 host.
64 warnings generated when compiling for gfx1102.
[ 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-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb32.cu:3:
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 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:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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 *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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(in/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
t); ++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-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-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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_tIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
 *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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<64, 2>' requested here
  704 |     flash_attn_comIn file included from bine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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<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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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, In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuhIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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
  505In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
 |     launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<DIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
  476 |         launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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 In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 hereIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |     ^

  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu:3:
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 gfx1103.
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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 gfx90a.
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-kqhalf-cpb16.cu/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu::33:
:
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.cuh/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh::11:
:
In file included from In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh::2020:
:
/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-hip/../ggml-common.h::171171::99::  warning: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]

    171171 |  |                 ssttrruucctt  {{

            | |         ^        ^

/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h9:: 192warning: :anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]9
: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192   | 192  |               s tsrturcutc t{ 
{      
|               ^| 
        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h: :warning: 213anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]:
9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |     213  |           s t r uscttr u{c
t       | {        ^

      |         ^
/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-hip/../ggml-common.h::254254::99::  warning: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]

    254254 |  |                 ssttrruucctt  {{

            | |         ^        ^

/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-hip/../ggml-common.h::281281::99::  warning: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]

    281281 |  |                 ssttrruucctt  {{

            | |         ^        ^

/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
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-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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) {
      |                                               ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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-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 In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_blocksIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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<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<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, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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 In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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_resulIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
ts<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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<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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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 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 |    In file included from  /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
    launch_fattn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_peIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |     ^
r_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<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<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<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<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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: In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_resIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
ults<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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, dsIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-kqfloat-cpb32.cu:3:
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);
      |     ^
64 warnings generated when compiling for gfx1150.
64 warnings generated when compiling for gfx1012.
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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-kqfloat-cpb32.cu:3:
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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.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-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
|                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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:
/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<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);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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 gfx1151.
64 warnings generated when compiling for gfx1030.
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:
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-kqfloat-cpb32.cu:3:
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);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh: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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &In file included from /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) {
      |                   ^
= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<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-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.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-kqfloat-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
61 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
64 warnings generated when compiling for gfx1152.
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-cpb8.cu:3:
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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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-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>
      In file included from |     ^
/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.cuhIn file included from :/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh476::29:
: /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuhnote: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here:
318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  476 |         launch_fatt  n318< | D,   p a rfaolrl e(li_nbtl ocksl = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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 |        In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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) {
      |                         ~ ^ ~~~~~~~~~~~
 *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:704:5: note: In file included from /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 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);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 gfx1010.
64 warnings generated when compiling for gfx1200.
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-kqhalf-cpb8.cu:3:
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-cpb16.cu:3:
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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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 |         laun/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
ch_fattn<D, parall/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
el_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-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) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
64 warnings generated when compiling for gfx1201.
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-kqhalf-cpb8.cu:3:
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-cpb16.cu:3:
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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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) {
      |                   ^
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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuhIn file included from :563:47: warning: function '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-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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:
/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<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);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 {
      |         ^
64 warnings generated when compiling for gfx900.
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-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb16.cu:3:
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 {
      |         ^
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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh  :554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 gfx906.
64 warnings generated when compiling for gfx1102.
58 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-kqhalf-cpb16.cu:3:
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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb8.cu:3:
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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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, paIn file included from ral/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cul:el3_:
b/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuhl:o14c:k35s:> (warning: cunused parameter 'Q' [-Wunused-parameter]t
x, dst, fattn   _14k | er n e l ,   n w acropnss,t  ccohlasr_ p*e r___brleosctkr,i cttr_u_e ,Q ,t
r      u| e                                  ^)
;
      |         ^
/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, 2>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 gfx908.
64 warnings generated when compiling for gfx1103.
58 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-kqhalf-cpb16.cu:3:
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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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:
/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 |       In file included from   /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cuc:o3n:
sIn file included from t/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh :i2n:
t/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh :n554b:2242:, 
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-cuda/template-instances/../fattn-wmma-f16.cuh:44:19: warning: unused parameter 'nb23' [-Wunused-parameter]
  554 |      44  |           * ( ( ucionnts3t2 _itn t* )n b&2K3Q,_
m      a| x                  ^_
sca/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuhl:e45): 19&:=  warning: funused parameter 'ne0' [-Wunused-parameter]t
z_mas   k45; | 
        |                         ^ 
    const in/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuht: 704n:e50:, 
note:       | in instantiation of function template specialization 'flash_attn_combine_results<128, 4>' requested here                  ^

/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:46:19  :704  | warning:  unused parameter 'ne1' [-Wunused-parameter] 
  fla   s46h | _ a t t n _ c o mcboinnset_ rienstu lntes1<,D
,       | p                  ^a
ralle/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuhl:_47:b19l:o cwarning: kunused parameter 'ne2' [-Wunused-parameter]s
>   
47 |         |      ^ 
   /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh: c476:o9ns:t note:  iin instantiation of function template specialization 'launch_fattn<128, 4>' requested heren
t ne2,  
476       |  |                    ^
   /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh: 48  :l19a: uwarning: ncunused parameter 'ne3' [-Wunused-parameter]h
_f   a48 | tt n  < D  ,   pacornalslt eli_ntb lnoeck3s)> {(
ct      | x                  ^,
 dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 gfx90a.
64 warnings generated when compiling for gfx1150.
58 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-kqhalf-cpb32.cu:3:
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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh: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<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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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<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 In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
/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<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_kernelIn file included from ,/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
 nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 |         *In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_resultsIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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,In file included from  /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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 |  In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
       *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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_combIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
ine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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_bIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
locks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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 In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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 |     flIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |     ^
ash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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, paraIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
llel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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 |     flaIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
sh_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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_blIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
ocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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;
      In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
|                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_In file included from blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_resuIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
lts<D, p/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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, parallelar_allel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |         ^
blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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 In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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, parallelIn file included from _blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 |         l_abulncohc_kfsa>t(tcnt<xD,,  dpsatr,a lflaetlt_nb_lokcekrsn>e(lc,t xn,w ardpsst,,  cfoaltst_np_ekre_rbnleolc,k ,n wtarrupes,,  tcroules)_;p
e      r| _        ^b
lock, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<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<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 gfx1102.
64 warnings generated when compiling for gfx90a.
64 warnings generated when compiling for gfx1151.
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/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
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.hIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
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 {
      |         ^
:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
  213 |         str/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
uct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an 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 {
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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;In file included from  l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
 < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh: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:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
  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_fattIn file included from /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]
   n_vec_cas14 |         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) {
      |                   ^
e(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-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_ma/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cux: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) {
      |                                  ^
_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
/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;
      |             ^~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<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<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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<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<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<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-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<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, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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;
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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<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<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<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<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-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<DIn file included from , /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cup:a3r:
aIn file included from l/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuhl:e2l:
_/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuhb:l554o:c24k:s >warning: 
cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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_s/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuhcale) &= ft:505:z_mas5: note: in instantiation of function template specialization 'launch_fattn<128, 1>' requested herek;
      |                        ^

  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: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<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<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<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-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<128, 4>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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<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<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<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;
      |                        ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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: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, coIn file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
ls_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<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<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<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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
In file included from /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/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned 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/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
*((uint32_t *) &KQ_max_scale) &= ft/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
z_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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/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,/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu: 126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]p
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
arallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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 |   /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48:   warning: suggest braces around initialization of subobject [-Wmissing-braces]lau
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
nch_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/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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/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/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |     ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:126:98: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  126 |         if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
58 warnings generated when compiling for gfx1103.
64 warnings generated when compiling for gfx942.
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-cpb8.cu:3:
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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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 In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
= 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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mas  704 |  k;
      |                        ^
   flash_attn_co/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
mbine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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:
/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<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-cpb16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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<128, 1>' requested here
  704 |     flash_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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<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<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,In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
 true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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<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<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<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<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 gfx1150.
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
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-cpb8.cu:3:
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/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-cpb32.cu:3:
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:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & mma_B) {
      |                                                                                                         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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/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) {
      |                                                                                                   ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & str      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
ide01, 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/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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, paralle/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
l_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:3:554:24: 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.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale)/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(const int D) {
      |                                               ^
 &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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:
/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/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t 2879 |             laun *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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);
      |         ^
ch_mul_mat_q<type,  80>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh_:t2691 :36*:)  warning: &comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]KQ_
max_scale) &=  2691f | t z _  m a s  k;i
f       (i| t                       ^ 
!= blockIdx.x || jt/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh !:= 704bl:o5c:k Inote: din instantiation of function template specialization 'flash_attn_combine_results<96, 2>' requested herex.
y) {
      |                                 ~~ ^  ~~~~~~~~~~  
704 |     fla/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhs:2805h:_9a:t tnote: nin instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 88, 8, false>' requested here
_combine_ 2805r |  e  s u l t s <mDu,l _pmaarta_lql_estlr_ebalmo_ckk_sf>i
xu      p<| t    ^y
pe, m/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuhm:q_491x:, 9M:M Qnote: in instantiation of function template specialization 'launch_fattn<96, 2>' requested here
_NWARPS, need_  ch491e |  c k >< < < b l olcaunkc_h_nfatutmsn_<xyD_t, ipalriangl,l ebll_obclockk_sd>i(mctsx,,  d0st,, s tfreaatmt>n>>_
k      | e        ^
rnel/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh,: 2882:n13w:a rnote: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 88>' requested herep
s, col s2882 | _ p e r _  bl o ck  ,   trluaeu,n cth_rumeu)l;_
m      a| t_        ^q
<type,  88>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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
In file included from  2813/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu | :  3 :
 In file included from  /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh : 2 :
mu/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuhl:_554m:a24t:_ qwarning: _cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]st
ream_k_fixup<type,   m554m | q _ x ,   M MQ _ N*WA(R(PuSi,n nte3e2d__tc he*c)k >&<K<Q<_bmlaoxc_ks_cnale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
ums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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/../fattn-common.cuh:704:5: note: in instantiation of function template specialization 'flash_attn_combine_results<128, 1>' requested here
  704 |     fl/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuha:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
sh_attn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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: /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
note: in instantiation of function template specialization 'launch_fattn<256, 4>' requested here
  476 |         launch_f/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
attn<D, parallel_blocks>(ctx, dst, fattn_kernel, nwarps, cols_per_block, true, true);
      |         ^
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |        /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
 *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 gfx1151.
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-cpb8.cu:3:
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 gfx1152.
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-cpb8.cu:3:
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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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 gfx1200.
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-cpb8.cu:3:
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_aIn file included from ttn_combine_results<D, parallel_blocks>
      |     ^
/builddir/build/BUILD/llama-cpp-b4580-build/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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu:3:
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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 gfx1201.
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-cpb8.cu:3:
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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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_m58 warnings generated when compiling for gfx900.
ax_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |     ^
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-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
58 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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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 {
      |         ^
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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  298 |         struct {
      |         ^
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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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 {
      |         ^
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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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);
      |     ^
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-cpb32.cu:3:
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 {
      |         ^
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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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/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-cpb32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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/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;
      |               ^~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
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/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
64 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/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-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 {
      |         ^
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/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) {
      |             ~~ ^  ~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu:3:
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.
[ 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 gfx1010.
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) {
      |             ~~ ^  ~~~~~~~~~~
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 {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
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 gfx1012.
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 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 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) {
      |             ~~ ^  ~~~~~~~~~~
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/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)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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-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) {
      |             ~~ ^  ~~~~~~~~~~
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/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 {
      |         ^
78 warnings generated when compiling for gfx1035.
/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/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 gfx1031.
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_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_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_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) {
      |             ~~ ^  ~~~~~~~~~~
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-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) {
      |             ~~ ^  ~~~~~~~~~~
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/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)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
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 {
      |         ^
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:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & 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) {
      |             ~~ ^  ~~~~~~~~~~
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 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 gfx1100.
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/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) {
      |             ~~ ^  ~~~~~~~~~~
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)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
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 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 gfx1103.
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.hIn 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 {
      |         ^
:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  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.cuhIn 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) {
      |                                                                                                         ^
:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & 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 * __restrIn file included from ict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/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:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^

/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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_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_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_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_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 (i/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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 | t   ! =   b l o c k I d x.lxa u|n| cjht_ m!u=l b_lmoactk_Iqd<xt.yy)p e{,
        | 3                                ~~ ^  ~~~~~~~~~~2
>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh::28132691::916::  note: warning: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ1_S, 24, 8, true>' requested herecomparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]

 2691 |         i f2813  | ( i t   ! =   b lmouclk_Imdaxt._xq _s|t|re ajmt_ k!_=f ibxluopc<ktIydpxe.,y )m m{q
_      x| ,            ~~ ^  ~~~~~~~~~~ 
MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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_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 != bloc/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhk:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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);
      Idx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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_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_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,/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh n:e2691e:d36_:c hwarning: eccomparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]k
><<<block_nums_xy_tiling ,2691  | bl o c k _ d i m si,f  0(,i ts t!r=e abml>o>c>k
I      d| x        ^.
x || jt /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh!:=2867 :b13l:o cnote: kin instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 48>' requested hereI
dx.y) {
      |                                 ~~ ^  ~~~~~~~~~~ 
2867 |             launc/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhh:_2805m:u9l:_ mnote: ain instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_IQ2_XS, 56, 8, false>' requested heret
_q<type,  48>(ctx,  2805a | r g s ,   s t r emauml)_;m
a      t| _            ^q
_stream_k_fixup/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh<:t2691y:p16:e ,warning:  comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]m
mq_x, MMQ_NWA R2691P | S ,   n e e d _ cihfe c(ki>t< <!<=b lbolcokc_knIudmxs._xx y|_|t ijlti n!g=,  bblloocckkI_ddxi.mys),  {0
,       | s            ~~ ^  ~~~~~~~~~~t
ream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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_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_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_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_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_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:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockId/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
x.y) {
 2691 |         if (it       |                                 ~~ ^  ~~~~~~~~~~
!/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh= blockIdx.x || jt != blo:2805:9:ckIdx.y) {
       |             ~~ ^  ~~~~~~~~~~
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_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: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_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: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_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_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_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_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_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_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_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_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_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) {
      |             ~~ ^  ~~~~~~~~~~
293 warnings generated when compiling for gfx90a.
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/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)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
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-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
   80 |     float tmp[n/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  cols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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 |             mul213_ |         struct {
      |         ^
mat_vec_q<type, 3><<<block_nums/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
, block_dims, 0, stream>>>(vx, vy, dst, 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-hip/../ggml-common.h:281:9: warning: anonymous types declared in 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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h::126298::989::  warning: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]

  298 |         struct {  
126       | |          ^ 
      if (threadIdx.x < rows_per_cuda_block && (rows_per_cuda_block == 1 || row0 + threadIdx.x < nrows_dst)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/mmvq.cu:80:48: warning: suggest braces around initialization of subobject [-Wmissing-braces]
   80 |     float tmp[ncols_y][rows_per_cuda_block] = {0.0f};
      |                                                ^~~~
      |                                                {   }
/builddir/build/BUILD/llama-cpp-b4580-build/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/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 gfx1150.
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_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 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) {
      |             ~~ ^  ~~~~~~~~~~
293 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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/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)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
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 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 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) {
      |             ~~ ^  ~~~~~~~~~~
293 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 {
      |         ^
/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)) {
      |                                                                               ~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
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, stream293 warnings generated when compiling for host.
>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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) {
      |             ~~ ^  ~~~~~~~~~~
[ 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
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
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_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_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-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 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 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 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 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 gfx1030.
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-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_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-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 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 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 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 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 gfx1200.
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 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 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 {
      |         ^
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_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;
      |               ^~~
78 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_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-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 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 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 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 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 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) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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_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-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_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_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 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 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 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 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 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.
[ 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 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 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, s78 warnings generated when compiling for gfx1010.
tream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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) {
      |             ~~ ^  ~~~~~~~~~~
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-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_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-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.
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.
[ 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 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 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 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.
[ 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 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 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 gfx1010.
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-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 gfx1035.
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 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 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 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 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 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 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 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 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 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 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 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 {
      |         ^
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-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 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 gfx1103.
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-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-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) {
      |             ~~ ^  ~~~~~~~~~~
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__ __for/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
ceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
_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: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 != 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) {
      |                                                                                                   ^
blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, 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_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:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     co/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
nst int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     con/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhst: 2691i:n36t:  nwarning: smcomparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare] 
  = ggml_cuda_info (2691) | . d e v i c e s [iifd ](.ints m!;=
       b| l              ^~~o
ckIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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-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-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-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-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 {
      |         ^
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:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & 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) {
      |             ~~ ^  ~~~~~~~~~~
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-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-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-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 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:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & 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 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 {
      |         ^
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:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & 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-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 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 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:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  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 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 {
      |         ^
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_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) {
      |             ~~ ^  ~~~~~~~~~~
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-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 gfx1010.
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 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 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 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 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 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 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 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 {
      |         ^
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-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 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 gfx900.
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-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-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, In file included from 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-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 |    /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
 __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & 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-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 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:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & 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-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.
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-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-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_muIn 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) {
      |                                                                                                         ^
l_mat_q<type, 112>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
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__ su/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
m, 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: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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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 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 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 {
      |         ^
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:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & 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) {
      |             ~~ ^  ~~~~~~~~~~
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-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.
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) {
      |                                                                                                   ^
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:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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-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 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 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_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-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 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 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 gfx1151.
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.
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) {
      |             ~~ ^  ~~~~~~~~~~
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-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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_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:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & 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_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_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.
[ 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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
78 warnings generated when compiling for gfx1152.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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 gfx942.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was 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]
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.
[ 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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was 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 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 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) {
      |             ~~ ^  ~~~~~~~~~~
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 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.
[ 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 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 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 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 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 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) {
      |             ~~ ^  ~~~~~~~~~~
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:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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-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-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-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-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-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-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 {
      |         ^
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:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & 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) {
      |             ~~ ^  ~~~~~~~~~~
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 gfx1102.
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;
      |               ^~~
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 {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
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-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-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-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-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-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-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 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 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 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 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 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.
[ 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) {
      |             ~~ ^  ~~~~~~~~~~
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 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) {
      |             ~~ ^  ~~~~~~~~~~
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 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 gfx1012.
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.
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-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-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;
      |               ^~~
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_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<t/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhype, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 :2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |   != bl   ockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
   if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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: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: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_ma/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuht_:q2691<:t36y:p ewarning: , comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare] 
56>(ctx, args, strea m2691) | ; 
        |              ^ 
   if (it /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh!:=2691 :b16l:o cwarning: kcomparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]I
dx.x || jt ! =2691  | b l o c k I d x .iyf)  ({i
t       | !                                ~~ ^  ~~~~~~~~~~=
 blockIdx.x |/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh|: 2813j:t9 :! =note:  in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q4_1, 16, 8, true>' requested hereb
lockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
 2813 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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_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_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_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_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_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_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_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:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
2691:36: /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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: /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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]
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) {
      |             ~~ ^  ~~~~~~~~~~
 2691 |         if (it != blockIdx.x || jt != blockIdx./builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhy) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh: 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,:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != bl mmq_x, MMQ_NWARPockIdSx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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
, nee 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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) {
      |             ~~ ^  ~~~~~~~~~~
d_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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_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_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: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_N/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhWA:RP2691S:,36 :n ewarning: ecomparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]d
_check><<<block_num s2691_ | x y _ t i l i n gi,f  b(liotc k!_=d ibmlso,c k0I,d xs.txr e|a|m >j>t> 
!      =|          ^b
lockId/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhx:.2894y:)13 :{ 
note:       in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 120>' requested here| 
                                ~~ ^  ~~~~~~~~~~
 2894 |     /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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 herel
aunch_mul_mat_q< t2813y | p e ,   1 2 0 > (mcutlx_,m aatr_gqs_,s tsrteraema_mk)_;f
i      x| u            ^p
<type, mm/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhq:_2691x:,16 :M Mwarning: Qcomparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]_
NWARPS, nee d2691_ | c h e c k > < < <ibfl o(cikt_ n!u=m sb_lxoyc_ktIidlxi.nxg ,| |b ljotc k!_=d ibmlso,c k0I,d xs.tyr)e a{m
>      >| >            ~~ ^  ~~~~~~~~~~

      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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_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_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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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 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 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-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 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 {
      |         ^
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-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-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_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-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-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-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 {
      |         ^
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:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & 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) {
      |             ~~ ^  ~~~~~~~~~~
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 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;
      |               ^~~
78 warnings generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
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.
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-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-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-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-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) {
      |                                                                                                   ^
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:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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: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_Q3_K, 8, 8, true>' requested here
 2813 |         mul_mat_q_stream_k_fixup<type, m/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
mq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 !=/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
 blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
 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_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_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_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_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_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_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_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_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_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_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_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_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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
 != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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: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:2691:36/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:: 2805:warning: 9:comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare] 
note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here
 2691  | 2805  |               i fm u(li_tm a!t=_ qb_lsotcrkeIadmx_.kx_ f|i|x ujpt< t!y=p eb,l omcmkqI_dxx,. yM)M Q{_
N      W| A                                ~~ ^  ~~~~~~~~~~R
PS, need_check/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh>:<2805<:<9b:l onote: cin instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 80, 8, false>' requested herek
_nums_xy_tiling,  2805b | l o c k _ d i m sm,u l0_,m astt_rqe_asmt>r>e>a
m      _| k        ^_
fixup<ty/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhp:e2879,: 13m:m qnote: _in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested herex
, MMQ_NWAR P2879S | ,   n e e d _ c h e c k >l<a<u<nbclho_cmku_ln_ummast__xqy<_ttyiplei,n g ,8 0b>l(occtkx_,d iamrsg,s ,0 ,s tsrteraema)m;>
>      >| 
            ^      
|         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691: 162879: |   warning:  comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare] 
         laun c2691h | _ m u l _ m a t _iqf< t(yipte ,! =  8b0l>o(ccktIxd,x .axr g|s|,  jstt r!e=a mb)l;o
c      k| I            ^d
x.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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: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_Q2_K, 88, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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: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:2691:36: 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/../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
 2691 |          2805i | f   ( i t   ! =  mbullo_cmkaItd_xq._xs t|r|e ajmt_ k!_=f ibxluopc<ktIydpxe.,y )m m{q
_      x| ,                                ~~ ^  ~~~~~~~~~~ 
MMQ_NWARPS, nee/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhd:_2805c:h9e:c knote: >in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, false>' requested here<
<<block_nums_xy_t i2805l | i n g ,   b l o cmku_ld_immast,_ q0_,s tsrteraema_mk>_>f>i
x      u| p        ^<
type, mmq/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh_:x2885,: 13M:M Qnote: _Nin instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested hereW
ARPS, nee d2885_ | c h e c k > < < < b l o clka_unnucmhs__mxuyl__tmialti_nqg<,t ybpleo,c k _9d6i>m(sc,t x0,,  asrtgrse,a ms>t>r>e
a      m| )        ^;

      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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 |  /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh : 2691 : 16 :   warning:  comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare] 
  launch_mul _2691m | a t _ q < t y p ei,f   (9i6t> (!c=t xb,l oacrkgIsd,x .sxt r|e|a mj)t; 
!      =|              ^b
lockIdx.y)/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh :{2691
:      16| :            ~~ ^  ~~~~~~~~~~ 
warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh::26912691::3636::  warning: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]

  26912691 |  |                 iiff  ((iitt  !!==  bblloocckkIIddxx..xx  ||||  jjtt  !!==  bblloocckkIIddxx..yy))  {{

            | |                                 ~~ ^  ~~~~~~~~~~                                ~~ ^  ~~~~~~~~~~

/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh::28132813::99::  note: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 96, 8, true>' requested herein instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 96, 8, true>' requested here

  28132813 |  |                 mmuull__mmaatt__qq__ssttrreeaamm__kk__ffiixxuupp<<ttyyppee,,  mmmmqq__xx,,  MMMMQQ__NNWWAARRPPSS,,  nneeeedd__cchheecckk>><<<<<<bblloocckk__nnuummss__xxyy__ttiilliinngg,,  bblloocckk__ddiimmss,,  00,,  ssttrreeaamm>>>>>>

            | |         ^        ^

/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh::28852885::1313::  note: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 96>' requested herein instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 96>' requested here

  28852885 |  |                         llaauunncchh__mmuull__mmaatt__qq<<ttyyppee,,    9966>>((ccttxx,,  aarrggss,,  ssttrreeaamm));;

            | |             ^            ^

/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh::26912691::1616::  warning: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]

  26912691 |  |                 iiff  ((iitt  !!==  bblloocckkIIddxx..xx  ||||  jjtt  !!==  bblloocckkIIddxx..yy))  {{

            | |             ~~ ^  ~~~~~~~~~~            ~~ ^  ~~~~~~~~~~

/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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: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_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: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:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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: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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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 != blo/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
ckIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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: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: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_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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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 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 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 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 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-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-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-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 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 gfx942.
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-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 {
      |         ^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:213:9: warning: /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.hanonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |  :192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
       struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h
:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  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-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-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) {
      |                                                                                                   ^
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:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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_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_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_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_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_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_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_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_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_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_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_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_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: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:2691:36: warning: /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhcomparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]:
2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  26912691 |  |                 iiff ( i(ti t! =! =b lbolcokcIkdIxd.xx. x| || |j tj t! =! =b lbolckoIcdkxI.dyx). y{)
       {| 
                                ~~ ^  ~~~~~~~~~~      
|             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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: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_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: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_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: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_Q2_K, 72, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhx,: 2691M:M36Q:_ Nwarning: WAcomparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]R
PS, need_check><<<block_ n2691u | m s _ x y _ t i liifn g(,i tb l!o= blockIdx.cxk _|d|i mjst,  !0=,  bsltorcekaImd>x>.>y
)       | {        ^

      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2805:9 :2876  | note:  in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, false>' requested here 
          launch_mu l2805_ | m a t _ q < t y pmeu,l _ m7a2t>_(qc_tsxt,r eaarmg_sk,_ fsitxruepa<mt)y;p
e      ,|              ^m
mq_x, MMQ_NWARP/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhS:,2691 :n16e:e dwarning: _comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]c
heck><<<blo ck2691_ | n u m s _ x y _ tiifl i(nigt,  !b=l obclko_cdkiImdsx,. x0 ,| |s tjrte a!m=> >b>l
o      c| k        ^I
dx.y) /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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./builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhx :|2691|: 36j:t  warning: !comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]=
 blockIdx.y) {
      |                                  ~~ ^  ~~~~~~~~~~2691
 |         if (it/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh :!2813=: 9b:l onote: cin instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 72, 8, true>' requested herek
Idx.x || jt != bl o2813c | k I d x . y )   {m
u      l| _                                ~~ ^  ~~~~~~~~~~m
at_q_stream_k/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh_:f2813i:x9u:p <note: tin instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q3_K, 80, 8, true>' requested herey
pe, mmq_x, MMQ_NWARPS, n e2813e | d _ c h e c k > <m<u<lb_lmoactk__qn_usmtsr_exaym__kt_ifliixnugp,< tbylpoec,k _mdmiqm_sx,,  0M,M Qs_tNrWeAaRmP>S>,> 
n      e| e        ^d
_check>/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh<:<2876<:b13l:o cnote: kin instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested here_
nums_xy_ti l2876i | n g ,   b l o c k _ d i mlsa,u n0c,h _smturle_amma>t>_>q
<      t| y        ^p
e,  72/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh>:(2879ctx,: 13a:r gnote: sin instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 80>' requested here,
 stream);
       | 2879            ^ | 
            la/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhu:n2691c:h16_:m uwarning: lcomparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]_
mat_q<type,  2691  | 8 0 > ( c t x ,  iafr g(si,t  s!t=r ebaml)o;c
k      | I            ^d
x.x || jt !/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh=: 2691b:l16o:c kwarning: Icomparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]d
x.y) {
      |             ~~ ^  ~~~~~~~~~~
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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: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_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: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_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_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_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_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_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_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]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691: 362691: |   warning:  comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare] 
     if (it != blockIdx.x | |2691  | jt  ! =   b l oc k Iidfx .(yi)t  {!
=       | b            ~~ ^  ~~~~~~~~~~l
ockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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: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><</builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh<:b2691l:o36c:k _warning: ncomparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]u
ms_xy_tiling,  b2691l | o c k _ d i m s ,i f0 ,( istt r!e=a mb>l>o>c
k      I| d        ^x
.x ||/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh :j2897t: 13!:=  note: bin instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 128>' requested herel
ockIdx.y )2897  | { 
        |                                  ~~ ^  ~~~~~~~~~~ 
       la/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhu:n2813c:h9_:m unote: lin instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 112, 8, true>' requested here_
mat_q<type, 128 >2813( | c t x ,   a r g sm,u ls_tmraeta_mq)_;s
t      r| e            ^a
m_k_fixup/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh<:t2691y:p16e:,  warning: mcomparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]m
q_x, MMQ_N W2691A | R P S ,   n e e di_fc h(eictk >!<=< <bblloocckkI_dnxu.mxs _|x|y _jtti l!i=n gb,l obclkoIcdkx_.dyi)m s{,
       0| ,            ~~ ^  ~~~~~~~~~~ 
stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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.
[ 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 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 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 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 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.
[ 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 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) {
      |                                                                                                   ^
78 warnings generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
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 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 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) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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 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 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 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 gfx1031.
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-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-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;
      |               ^~~
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) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
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 /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhint & k00) {
      |                                                                                                   ^
:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, 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_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 |       /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh  m:ul_1704m:a99:t _warning: qunused parameter 'k00' [-Wunused-parameter]_
stream_k_fixup<ty p1704e | ,   m m qc_oxn,s tM MiQn_tN W*A R_P_Sr,e sntereidc_tc_h_e cxk,> <c<o<nbslto cikn_tn u*m s___xrye_sttirliicntg_,_  byl,o cfkl_odaitm s*,  _0_,r essttrreiacmt>_>_> 
s      u| m        ^,
 const/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh :i2858n:t13 :&  note: kin instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 24>' requested here0
0) {
      |                                                                                                   ^ 
2858 |             launch_mul_mat_q<type,  24>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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./builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & x || jtne00, const int & ne01,  != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
c/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdxonst int & stride01,. consxt int & ne1 || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
0, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_tili78 warnings generated when compiling for gfx908.
ng, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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_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_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 != blockId/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
x.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
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 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 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 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 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 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 {
      |         ^
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-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 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 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;
      |               ^~~
78 warnings generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
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, strea78 warnings generated when compiling for host.
m>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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) {
      |             ~~ ^  ~~~~~~~~~~
[ 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
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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-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 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) {
      |             ~~ ^  ~~~~~~~~~~
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-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.
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_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_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 != blockIdIn file included from x.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_/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__ voidc hmemcak_><<K<8b(lcoocnks_tn ummmas__ixnyt__tAi_lIi1n6gK,8  b&l omcmka__Ad,i mcosn,s t0 ,m msat_rienatm_>B>_>J
8      K| 8        ^ 
& mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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) {
      |             ~~ ^  ~~~~~~~~~~
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 i/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhn:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
t * __restrict__ y, floa/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
t * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, 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_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:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
78/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
 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_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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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;
      |               ^~~
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:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
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 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 gfx1151.
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-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 gfx1151.
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-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-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-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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
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:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, 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: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:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restr/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
ict__ 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: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:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was 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 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) {
      |             ~~ ^  ~~~~~~~~~~
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 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 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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was 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 gfx1152.
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 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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was 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]
78 warnings generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was 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-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 {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was 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-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:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported 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_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:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported 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_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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Wpass-failed=transform-warning]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was 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 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 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 != blo134 warnings generated when compiling for gfx1201.
ckIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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-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 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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was 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]
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_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 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 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-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) {
      |             ~~ ^  ~~~~~~~~~~
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 gfx1200.
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_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_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:2691In 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) {
      |                                                                                                         ^
:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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, MIn 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) {
      |                                                                                                   ^
MQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, 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_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:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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_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_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_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_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_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:2805:9: note: in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_0, 32, 8, false>' requested here/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh
:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2805 |         mul_ m2691a | t _ q _ s t r e aimf_ k(_ifti x!u=p <btlyopcek,I dmxm.qx_ x|,|  MjMtQ _!N=W AbRlPoSc,k Indexe.dy_)c h{e
c      k| >                                ~~ ^  ~~~~~~~~~~<
<<block_nums_x/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhy:_2813t:i9l:i nnote: gin instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q5_1, 128, 8, true>' requested here,
 block_dims, 0,  s2813t | r e a m > > > 
       m| u        ^l
_mat_/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhq:_2861s:t13r:e anote: min instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 32>' requested here_
k_fixup<t y2861p | e ,   m m q _ x ,   M M Ql_aNuWnAcRhP_Sm,u ln_emeadt__cqh<etcykp>e<,< < b3l2o>c(kc_tnxu,m sa_rxgys_,t isltirnega,m )b;l
o      c| k            ^_
dims, 0, /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhs:t2691r:e16a:m >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/../mmq.cuh :26912897 | : 13 :   note:  in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 128>' requested here 
  if (it  2897! | =   b l o c k I d x . x  l|a|u njcth _!m=u lb_lmoactk_Iqd<xt.yyp)e ,{ 
1      2| 8            ~~ ^  ~~~~~~~~~~>
(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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 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 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.
[ 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 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 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 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-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 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 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 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 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 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 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 {
      |         ^
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:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & 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) {
      |             ~~ ^  ~~~~~~~~~~
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_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 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:78 warnings generated when compiling for gfx1031.
2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
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-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-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 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 gfx908.
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-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-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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
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:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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 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 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 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 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 {
      |         ^
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:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & 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) {
      |             ~~ ^  ~~~~~~~~~~
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.
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-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-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;
      |               ^~~
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_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: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_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: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: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_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: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]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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: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_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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:2691::36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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: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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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
_stream_k_fixup< type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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) {
      |             ~~ ^  ~~~~~~~~~~
2805 |         mul_mat_q_stream_k_fixup<t/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
ype, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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: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_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: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 != blo/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
ckIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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);_stream_k_fixup<t
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
ype, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != bloc warning: kIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare] 2691 |         if (it !
= blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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 |            2691 |         if   launch_mul_mat_q<type,  64>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockId(ix.y) {
      |             ~~ ^  ~~~~~~~~~~
t != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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                                ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 | m       at_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh: if (it2876: != bl13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 72>' requested here
 2876 |      ockIdx.  x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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
     launch_mul_mat_q <2805t | y p e ,     7 2 >m(uclt_xm,a ta_rqg_ss,t rsetarme_akm_)f;i
xu      p| <            ^t
ype, mmq_x, MMQ_NWARPS, need_chec/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhk:>2691<:<16<:b lwarning: ocomparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]c
k_nums_xy_tiling ,2691  | b l o c k _ d i misf,  (0i,t  s!t=r ebalmo>c>k>I
d      x| .        ^x
 || jt /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh!:=2894 :b13l:o cnote: kin instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 120>' requested hereI
dx.y) {
      |             ~~ ^  ~~~~~~~~~~ 
2894 |             launch_mul_mat_q<type, 120>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it !=/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
 blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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: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_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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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-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 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 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 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 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 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.
[ 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 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 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 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 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 gfx1012.
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-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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1022:5: warning: loop not unrolled: the optimizer was 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 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 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) {
      |             ~~ ^  ~~~~~~~~~~
134 warnings generated when compiling for gfx942.
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-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-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-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_J8KIn 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) {
      |                                                                                                   ^
4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with 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:1053:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1053 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
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:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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: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_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: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_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: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_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: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:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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: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_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: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_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: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]
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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: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_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: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_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
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh :2867 | 2691 : 36 :   warning:   comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare] 
    launch_mul_ m2691a | t _ q < t y p e ,i f  4(8i>t( c!t=x ,b laorcgksI,d xs.txr e|a|m )j;t
       !| =            ^ 
blockIdx.y/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh): 2691{:
16      :|                                  ~~ ^  ~~~~~~~~~~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/../mmq.cuh:2813 :26919 | :   note:  in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q2_K, 56, 8, true>' requested here 
    if (it != bl o2813c | k Id x . x   | |  mjutl _!m=a tb_lqo_cstkrIedaxm._yk)_ f{i
x      u| p            ~~ ^  ~~~~~~~~~~<
type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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_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:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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>/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh<:<<2805b:l9o:c knote: _in instantiation of function template specialization 'mul_mat_q_stream_k_fixup<GGML_TYPE_Q8_0, 64, 8, false>' requested heren
ums_xy_tiling, block _2805d | i m s ,   0 ,   smturle_amma>t>_>q
_      s| t        ^r
eam_k_f/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhi:x2876u:p13<:t ynote: pin instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 72>' requested heree
, mmq_x, M M2876Q | _ N W A R P S ,   n e e dl_acuhnecchk_>m<u<l<_bmlaotc_kq_<ntuympse_,x y _7t2i>l(icntgx,,  abrlgosc,k _sdtirmesa,m )0;,
       s| t            ^r
eam>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh::162873:: 13warning: :comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare] 
note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 64>' requested here
  26912873 |  |                 i f   ( ilta u!n=c hb_lmouclk_Imdaxt._xq <|t|y pjet,  ! =6 4b>l(occtkxI,d xa.ryg)s ,{ 
s      t| r            ~~ ^  ~~~~~~~~~~e
am);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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_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_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_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_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_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_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 != /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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_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_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:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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 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 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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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 gfx1152.
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-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-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;
      |               ^~~
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_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_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_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_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_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:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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:2885:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 96>' requested here
 2885 |         /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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]
     2691l | a u n c h _ m u li_fm a(ti_tq <!t=y pbel,o c k9I6d>x(.cxt x|,|  ajrtg s!,=  sbtlroecakmI)d;x
.      y| )            ^ 
{
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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_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: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_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_Q6_K, 112, 8, false>' requested here
 2805 |         mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_x/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhy:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
_tiling, block_dim/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuhs:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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: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_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 !/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
= blockIdx.x || jt/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
 != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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 |             laun/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
ch_mul_mat_q<type, 1/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
20>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:16: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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 |  /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
       mul_mat_q_stream_k_fixup<type, mmq_x, MMQ_NWARPS, need_check><<<block_nums_xy_tiling, block_dims, 0, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_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:/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
 warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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 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-q4_0-q4_0.cu:3:
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.
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/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_casIn 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 {
      |         ^
e_impl<D, cols_per_block,/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
 parallel_blocks, type_/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
K, type_V, use_lo/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
git_softcap>(ctx, dst);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(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-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  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-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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
60 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_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) {
      |             ~~ ^  ~~~~~~~~~~
In 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-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 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) {
      |             ~~ ^  ~~~~~~~~~~
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-q4_0-q4_0.cu:3:
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 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 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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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 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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
78 warnings generated when compiling for gfx908.
/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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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 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-hs128-q4_0-q4_0.cu:3:
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 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 {
      |         ^
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:5:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:172:105: warning: function 'mma_K4' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  172 |     __device__ __forceinline__ void mma_K4(const mma_int_A_I16K4 & mma_A, const mma_int_B_J8K4 & mma_B) {
      |                                                                                                         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mma.cuh:194:105: warning: function 'mma_K8' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  194 |     __device__ __forceinline__ void mma_K8(const mma_int_A_I16K8 & mma_A, const mma_int_B_J8K8 & 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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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 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 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-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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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 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 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-q4_0-q4_0.cu:3:
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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
60 warnings generated when compiling for gfx1035.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
In 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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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.
[ 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 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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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 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 gfx1102.
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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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-q8_0-q8_0.cu:3:
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 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 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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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 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 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 |         s60 warnings generated when compiling for gfx1151.
truct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In 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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
78 warnings generated when compiling for gfx90a.
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) {
      |             ~~ ^  ~~~~~~~~~~
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 {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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 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 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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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 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 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 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-q4_0-q4_0.cu:3:
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 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 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-q4_0-q4_0.cu:3:
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 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 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 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 gfx1201.
78 warnings generated when compiling for gfx942.
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-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/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-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In 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, 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-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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-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);
      |         ^
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) {
      |                                                                                                         ^
/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/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/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-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);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-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/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_bloc/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
ks, type_K, type_V, use_logit_softcap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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 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 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 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 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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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 {
      |         ^
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/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) {
      |             ~~ ^  ~~~~~~~~~~
In 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 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 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 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 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 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.
[ 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 gfx90a.
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-q8_0-q8_0.cu:3:
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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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) {
      |                   ^
In 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-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-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-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-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_K/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
Q_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-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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_v/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                              ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
ec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_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 |   /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  |                  f o r   (gignmtl _ic0u d=a _0f;l ais0h _<a tDt/ns_iezxeto_fv(eicn_tf)1;6 _ic0a s+e=_ iWmApRlP<_DS,I ZcEo)l s{_
p      e| r                                 ~~ ^ ~~~~~~~~~~~~~_
block, parallel_blocks, type/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh_:K284,: 35t:y pnote: ein instantiation of function template specialization 'flash_attn_vec_ext_f32<128, 4, 4, GGML_TYPE_Q8_0, GGML_TYPE_Q8_0, true>' requested here_
V, use_logit_softcap  >284( | c t x ,  fdasttt)n;_
k      e| r            ^n
el_t fattn_ke/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuhr:n129e:l33 :=  warning: fcomparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]l
ash_attn_vec_e  x129t | _ f 3 2 < D ,   c o l s _fpoerr _(bilnotc ki,0  p=a r0a;l lie0l _<b lDo/cskisz,e otfy(pien_tK),;  tiy0p e+_=V ,W AuRsPe__SlIoZgEi)t _{s
o      f| t                             ~~ ^ ~~~~~~~~~~~~~c
ap>;
      |                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuh:142:33: /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuhwarning: :comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]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
  142 |               337f | o r   ( i n t   i 0   =  g0g;m li_0c u<d aD_/fsliazseho_fa(titnnt_)e;x ti_0v e+c=_ fW3A2R_Pc_aSsIeZ_Ei)m p{l
<      D| ,                             ~~ ^ ~~~~~~~~~~~~~ 
cols_per_block, parallel_blocks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f32.cuh:129:33: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  129 |             for (int i0 = 0; i0 < D/sizeof(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_kerne/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuhl_t fattn_kernel = flash_attn_vec_ext_f32<D, cols_per_block, parallel_blocks, type_K, type_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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
In 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:116:37: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  116 |                 for (int i0 = 0; i0 < D/sizeof(int); i0 += WARP_SIZE) {
      |                                  ~~ ^ ~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/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) {
      |                              ~~ ^ ~~~~~~~~~~~~~
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) {
      |             ~~ ^  ~~~~~~~~~~
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-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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) {
      |                         ~ ^ ~~~~~~~~~~~
60 warnings generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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-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.
[ 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
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-hs256-f16-f16.cu:3:
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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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-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,In file included from  c/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cuo:l3s:
In file included from _/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-vec-f16.cuhp:e2r:
_b/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuhl:o554c:k24,:  pwarning: arcast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]a
llel_blocks, ty  p554e | _ K ,   t y p e _*V(,( uuisnet_3l2o_gti t*_)s o&fKtQc_ampa>x(_cstcxa,l ed)s t&)=; 
f      t| z            ^_
mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-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 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 {
      |         ^
In 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 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-hs256-f16-f16.cu:3:
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, paral60l warningse generatedl_b when compiling for lgfx1100o.
cks, type_K, type_V, use_logit_softcap>(ctx, dst);
      |             ^
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);
      |         ^
In 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);
      |         ^
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 {
      |         ^
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/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);
      |         ^
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) {
      |                                                                                                   ^
60 warnings generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1704:99: warning: unused parameter 'k00' [-Wunused-parameter]
 1704 |     const int * __restrict__ x, const int * __restrict__ y, float * __restrict__ sum, const int & k00) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:17: warning: unused parameter 'ne00' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                 ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2497:75: warning: unused parameter 'ne10' [-Wunused-parameter]
 2497 |     const int & ne00, const int & ne01, const int & stride01, const int & ne10, const int & ne11, const int & stride11, const int & ne0,
      |                                                                           ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2821:15: warning: unused variable 'nsm' [-Wunused-variable]
 2821 |     const int nsm   = ggml_cuda_info().devices[id].nsm;
      |               ^~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2691:36: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
 2691 |         if (it != blockIdx.x || jt != blockIdx.y) {
      |                                 ~~ ^  ~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-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) {
      |             ~~ ^  ~~~~~~~~~~
In 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);
      |         ^
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);
      |         ^
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 {
      |         ^
In 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);
      |         ^
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);
      |         ^
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-q8_0-q8_0.cu:3:
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) {
      |                         ~ ^ ~~~~~~~~~~~
34 warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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 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-hs256-f16-f16.cu:3:
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 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 {
      |         ^
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-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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 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-hs256-f16-f16.cu:3:
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-hs128-f16-f16.cu:3:
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 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);
      |         ^
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) {
      |             ~~ ^  ~~~~~~~~~~
34 warnings generated when compiling for gfx1103.
78 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
In 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-hs64-f16-f16.cu:3:
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: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-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);
      |         ^
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 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 {
      |         ^
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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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 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);
      |             ^
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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-hs64-f16-f16.cu:3:
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);
      |         ^
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 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) {
      |                                               ^
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:
/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 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);
      |             ^
34In file included from  warnings generated when compiling for gfx1031.
/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-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);
      |         ^
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 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_TYP34 warnings generated when compiling for gfx1150.
E_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 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-f16-f16.cu:3:
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 gfx1100.
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-hs64-f16-f16.cu:3:
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-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-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-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.
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-hs64-f16-f16.cu:3:
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);
      |         ^
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.
In 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 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-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,
In file included from       |                   ^
/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-instance-hs128-q8_0-q8_0.cu:3:
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-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-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 gfx1152.
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-hs64-f16-f16.cu:3:
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: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);
      |         ^
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 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);
      |         ^
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);
      |             ^
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-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-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-hs256-f16-f16.cu:3:
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 {
      |         ^
34 warnings generated when compiling for gfx1151.
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:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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);
      |         ^
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-hs128-f16-f16.cu:3:
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-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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);
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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.
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 gfx906.
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 {
      |         ^
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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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-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 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: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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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-hs256-f16-f16.cu:3:
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-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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: 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) {
      |                   ^
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: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 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 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 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 {
      |         ^
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:2:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:318:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  318 |     for (int l = 0; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:325:23: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  325 |     for (int l = 1; l < sizeof(int); ++l) {
      |                     ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:341:27: warning: comparison of integers of different signs: 'int' and 'unsigned long' [-Wsign-compare]
  341 |         for (int l = 0; l < sizeof(int); ++l) {
      |                         ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563 | static void on_no_fattn_vec_case(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);
      |         ^
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 {
      |         ^
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-q8_0-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-hs64-f16-f16.cu:3:
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 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-hs64-f16-f16.cu:3:
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:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-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-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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 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 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);
      |             ^
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-hs64-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-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.
[ 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 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 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 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 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.
[ 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 gfx1010.
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-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-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);
      |         ^
In 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.
[ 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 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 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 gfx1012.
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-hs256-f16-f16.cu:3:
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);
      |         ^
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 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-hs64-f16-f16.cu:3:
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-hs128-f16-f16.cu:3:
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.
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);
      |         ^
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-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.
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.
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-hs64-f16-f16.cu:3:
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-hs128-f16-f16.cu:3:
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 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 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 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 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 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 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 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 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-hs64-f16-f16.cu:3:
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 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-f16-f16.cu:3:
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-hs64-f16-f16.cu:3:
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-hs128-f16-f16.cu:3:
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.
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-hs64-f16-f16.cu:3:
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-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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: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);
      |             ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:554:24: warning: cast from 'const float *' to 'unsigned int *' drops const qualifier [-Wcast-qual]
  554 |         *((uint32_t *) &KQ_max_scale) &= ftz_mask;
      |                        ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../fattn-common.cuh:563:47: warning: function 'on_no_fattn_vec_case' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  563In 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);
      |             ^
 | static void on_no_fattn_vec_case(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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-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:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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 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.
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 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 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 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 {
      |         ^
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-hs64-f16-f16.cu:3:
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-hs128-f16-f16.cu:3:
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-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-hs128-f16-f16.cu:3:
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 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 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 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.
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 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-f16-f16.cu:3:
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 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 gfx1151.
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-hs256-f16-f16.cu:3:
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:1:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../common.cuh:20:
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:171:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  171 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:192:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  192 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:213:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  213 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:254:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  254 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:281:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-types]
  281 |         struct {
      |         ^
/builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-hip/../ggml-common.h:298:9: warning: anonymous types declared in an anonymous union are an extension [-Wnested-anon-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-hs128-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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-hs256-f16-f16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b4580-build/llama.cpp-b4580/ggml/src/ggml-cuda/template-instances/../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: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-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 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 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 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 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 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 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-hs256-f16-f16.cu:3:
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.
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 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 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 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 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.
[ 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
[ 78%] Building CXX object src/CMakeFiles/llama.dir/llama-batch.cpp.o
[ 79%] Building CXX object src/CMakeFiles/llama.dir/llama-adapter.cpp.o
[ 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.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
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
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
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
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
[ 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
[ 94%] Building CXX object common/CMakeFiles/common.dir/console.cpp.o
[ 95%] Building CXX object common/CMakeFiles/common.dir/common.cpp.o
[ 94%] 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/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
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
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
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
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
[ 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.0FMmbM
+ 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 -j4 --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 -j4
+ /usr/lib/rpm/redhat/brp-python-hardlink
+ /usr/bin/add-determinism --brp -j4 /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.B9NZz3
+ 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.PEVaQb
+ 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) = 1b69a95b5a29d9edbc7bdba24b2114ddfd864a76 debuginfo(build-id) = 504abc4b05fd73f0a88d650642c426a4fbb2ed89 debuginfo(build-id) = 8ee063b29dd4cc8ceb5b12443443fc6055fa2485 debuginfo(build-id) = a73dc9166d815caf5f75f90a88bd52ba7a9e20e8 debuginfo(build-id) = bbd83d6ac6d57606d1a593e996c4b5987827bb3c 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-devel-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-debuginfo-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.RB6q7z
+ 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_64-1741555222.965739/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) 92 minutes 20 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": "src"
        },
        {
            "name": "llama-cpp-devel",
            "epoch": null,
            "version": "b4580",
            "release": "2.fc43",
            "arch": "x86_64"
        },
        {
            "name": "llama-cpp-debugsource",
            "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",
            "epoch": null,
            "version": "b4580",
            "release": "2.fc43",
            "arch": "x86_64"
        }
    ]
}
RPMResults finished